Home

Submission Docs

1. ARIZONA STATE UNIVERSITY Department of Electrical Engineering Arizona State University Tempe AZ 85287 7206 Tel 480 965 5311 Fax 480 965 8325 Email spanias asu edu 20 July 2003 2003 Premier Award c o NEEDS 3115 Etcheverry Hall University of California at Berkeley Berkeley CA 94720 1750 Dear Sir Madam I respectfully submit my engineering education software entitled Java Digital Signal Processing J DSP for consideration for the 2003 Premier award Enclosed are 15 copies of the J DSP submission packet We refer to J DSP as abware because it enables student users to perform online simulations and computer experiments over the Internet The J DSP concept is due to Professor Andreas Spanias who is also the chief software designer and the main author of the labware Collaborators include graduate students who have assisted the main author with various software tasks and colleagues who used the software in their classes The J DSP project is sponsored by Arizona State University and by NSF CCLI EMD 0089075 The copyright owner of the J DSP source and executable code is the Arizona State Board of Regents that is by law holder of all intellectual property generated by Arizona State University It is a non profit project and therefore J DSP is freely accessible over the Internet not only to ASU students but also to students at other Universities and DSP practitioners J DSP executables ISBN 0 9724984 0 0 have already
2. Periodi Samples Time Shif fo Close Update Help ja Unsigned Java Applet Window Fig 5 Signal generator dialog On the right side of the signal generator window you can see a preview of the signal You may change the name of the signal the gain the pulse width the period and the time shift by typing the desired value into the appropriate box The signal type can be changed by clicking on the drop down menu and selecting a signal If you select a User defined signal an Edit signal button will appear allowing you to edit the signal With all signals except audio J DSP assumes a normalized sampling frequency of 1Hz Hence the sampling frequency in terms of radians is 27 All frequencies are entered as a function of 7 e g 0 1m 0 3567 etc Any sinusoidal frequency at or above m will result in aliasing Step 1 1 Create a sinusoid with frequency 0 17 amplitude 3 75 pulse width 40 When all of the parameters have been entered press the Update button to update the signal preview Remember that whenever changes are made to this box the Update button must be pressed in order for the changes to take effect On the right you can see a preview of the input signal Count the number of samples within a period How many do you have ans 20 samples M1 5 Step 1 2 Create a sinusoid with frequency 7 amplitude 3 75 pulse width 40 remember to press u
3. 10 Ko Y Duman T Spanias A J DSP for Communications 33rd ASEE ITEEE FIE 03 Boulder Nov 2003 11 Spanias A et al On Line Laboratories for Speech and Image Processing and for Communication Systems Using J DSP IEEE 2002 DSP Workshop Callaway Georgia October 2002 10 FROM Panasonic FAX SYSTEM PHONE NO E Jun 07 2003 05 16PM PS YE Northeastern U N I V ER SIT Y College of kanaa To The NEEDS award committee Department of Electrical and From Professor John Proakis Computer Engineering Date 6 4 03 440 Dono Research Cenlor Subject Recommendation letter of the J DSP educational software Northeastern University ics Boston Massechuserts 02115 5000 Phone 617 373 4159 Facsimile 617 373 8970 This is regarding the educational Java software J DSP that was developed by Prof Andreas Spanias and his team at Arizona State University I became aware of J DSP at the 2002 IEEE ASEE Frontiers in Education conference in Boston where J DSP was freely disseminated to Engineering educators received a free copy of this software for evaluation The J DSP software was developed solely for instruction at Arizona State University with funds from the state of Arizona and the National Science Foundation The objective of the original J DSP project was to develop a programming environment to provide computer laboratory experiences over the internet to distance learners The outcome of this p
4. 3 value B3 plot2 3 1 gt M2 12 M2 9 Block name Sound Player Notation SndPlyr Description This block is used for signal playback Dragging the volume scroll bar to the right increases the signal volume Pin assignment Input signal x n Dialog window s Sound Player fe x 9 0E 1 0 0 T 0E 1 0 5192 Java Applet Window a Sound Plyr dialog window Script use Name sndplayer Example code lt param name 3 value B3 sndplayer 3 1 gt M2 13 M2 10 Block name Quantizer Notation Quantizer Description This block is used for signal quantization Uniform or non uniform quantization can be selected For uniform quantization the amplitude levels are divided into steps of 0 5 where n is the number of quantization bits These discrete levels are used to represent the signal amplitudes Non uniform quantization is achieved by uniformly quantizing a U law or A law compressed signal Note that this block can only simulate the effect of quantization on signals or on filter coefficients Pin assignment Quantized signal y n Dialog window s M Quantizer f 4 x LiUantization Uniform Mon unitormnu lawiA laywy Uniform Quantization C Non Uniform Quantization Mumber of Quantization bits e Close Update Java Applet Window a Quantizer dialog window Script use Name quant Example code lt param name 3 value B3 qua
5. 1996 Award from Intel Corporation Portland In Appreciation of Support for the Intel Research Program 1993 Award from Intel Corporation for Leadership and Contributions in the Development of the Intel 60172 Signal Processing Architecture 1992 Team Award from Intel Corporation to Andreas Spanias and three of his graduate students for contributions in the development of speech processing algorithms 1983 Rufus West Achievement Award at WVU e Member Sigma Xi Tau Beta Pi HKN Golden Key Publications Refereed Archival Journal Papers a Published or Accepted for Publication J 1 Ted Painter and A Spanias Sinusoidal Analysis Synthesis of Audio using Perceptual Criteria To appear in the IEEE Transactions on Speech and Audio J 2 Gopal Nair and A Spanias The Eigenspace Projection Algorithm Signal Processing Accepted and will appear in 2003 J 3 J Foutz A Spanias S Bellofiore and C Balanis Adaptive Eigen Projection Beamforming Algorithms for 1 D and 2 D Antenna Arrays Accepted in IEEE Antennas and Propagation Letters 2003 J 4 Ted Painter and Andreas Spanias Sinusoidal Analysis Synthesis of Audio Using Perceptual Criteria EURASIP JASP Special Issue On Multimedia Signal Processing Vol 2003 Issue 1 pp 15 20 January 2003 J 5 S Bellofiore J Foutz C Balanis A S Spanias T Duman J Capone Smart Antennas for Mobile adhoc Networks IEEE Trans on Antennas and Propagation pp 571 581 Vol
6. 4 Close Update Help Warning Applet window Abways press update to initiate a change Note that the filter coefficients correspond to the following diference equation i i ye gt ax e D Dayle i iml i l Fig 6 Coefficient entry in J DSP M1 6 Step 2 2 Keep the values in SigGen as per step 1 1 Change the filter coefficient to b0 4 and press Update Double click on the Plot block You should see that the amplitude of the sinusoid has changed ans peak amplitude 4x3 75 15 Step 2 3 Implement a pure delay by setting b5 1 and rest of the coefficients including bO to zero and press Update What happens to the sinusoid Step 2 4 Implement a simple LPF set bO 0 2 and al 0 8 and press Update Generate a sinusoid with gain 1 frequency 0 17 pulse width 256 What do you observe What kind of signal do you get at the output Why What is the peak to peak value Do we have a change Is there a phase shift What filter function determines the time shift Step 2 5 Select the Freq Resp block from the panel of general blocks on the left of the window and place it to the north of the Filter block Connect the parameter output to the Freq Resp block Double click the Freg Resp block You should see the magnitude and phase response of the filter Change the coefficient to al 0 8 instead of al 0 8 What do you see in the frequency response and output ans HPF decrease in amplitude
7. A new phase model for sinusoidal transform coding IEEE Trans on Speech and Audio Processing vol 6 no 5 pp 495 501 Sept 1998 A Spanias principal investigator and Ph D advisor 10 M Deisher and A S Spanias Speech enhancement using state based estimation and sinusoidal modeling Journal of Acoustical Society of America vol 102 2 pp 1141 1148 Aug 1997 A Spanias principal investigator and Ph D advisor 11 P Loizou and A Spanias Improving discrimination of confusable words using the divergence measure Journal of Acoustical Society of America Vol 101 2 pp 1106 1111 Feb 1997 A Spanias principal investigator and Ph D advisor 12 P Loizou and A S Spanias High Performance Alphabet Recognition IEEE Trans on Speech and Audio Processing vol 4 no 6 pp 430 445 Nov 1996 A Spanias principal investigator and Ph D advisor 13 K Tsakalis M Deisher and A S Spanias System identification based on bounded error constraints IEEE Trans on Signal Processing vol 43 no 12 pp 3071 3075 Dec 1995 A Spanias co principal investigator and Ph D advisor 14 M Deisher and A S Spanias Practical considerations in the implementation frequency domain adaptive noise cancellation IEEE Trans on Circuits and Systems Part II Analog and Digital Signal Processing vol 41 no 2 pp 164 168 Feb 1994 A Spanias principal investigator and Ph D advisor 15 A S
8. k Liy R k ph w n window x n The p frame of the time domain input signal R k Welch PSD estimate of all the frames R k Sample PSD estimate of the p frame M7 8 M7 8 Block name Spectrogram Notation Spectrogram Description This block calculates the spectrogram frequency versus time plot of the given input signal The window types available are Hamming Hanning rectangular Gaussian Bartlett and Kaiser The window length the number of FFT points and the resolution can be specified by the user By moving the cursor on the plot the normalized magnitude and the xy coordinates can be viewed Pin assignment Time domain signal x n 6 Dialog window s Spectrogram block ii x Window Type Window Length 128 Mo of FFT points 256 Resolution z f dB C Linear normalized frequency 0 5 0 4096 5191 time samples Close Update Help Java Applet Window a Spectrogram dialog window Script use Name specgram Example code lt param name 3 value B2 specgram 3 1 gt M7 9 M7 10 Block name AR estimator Notation AR Est Description This block computes the AR coefficients and plots the auto regressive spectrum of the input signal using the Levinson Durbin algorithm The following lag windows are available rectangular Hamming triangular and Gaussian The maximum number of AR coefficients allowed 64 Pin assignment Tim
9. Filter type can be low pass high pass or pass band Wp Ws pass band and stop band edge cut off frequencies respectively Wp2 Ws2 second pass band and stop band edge cut off frequencies respectively for pass band filters PB SB pass band and stop band tolerances in dB Cut off frequences f e take values from 0 to 1 where fe 1 corresponds to half the sampling frequency The design process is illustrated in terms of the block diagram below Discrete time Map Calculate low Transform Apply Calculate filter discrete pass filter to desired bilinear coefficients specification specs to prototype type low transform continuous Butterworth pass high time Chebyshev or bass band Elliptic pass Pin assignment escription Filter coefficients 2 Ea Dialog window s IIR Filter Parameters Mame p IIR filter Eliptic zj JumCoet DenCoeft Pass Band 0 0935 1 0 0 4 1 0 0 298 0 1365 Fiter ype HighPass n486 11107 Order 4 0 258 0 3148 Cut off Frequencies 0 0935 0 4779 wot a4 Wel 0 2 0 0 00 Wipe 0 5 Wis ar 0 0 0 0 UU 0 0 Tolerances dB 0 0 0 0 FE 5 0 SB 50 0 Close Update Hep warning Applet window a FIR dialog window Script use Name IIR Example code lt param name 3 value B3 ITR 3 1 gt M6 5 M6 5 Block name Kaiser design Notation Kaiser Description This block designs Kaiser FIR filters based on the windowing m
10. Haresh Corba E0810 F dapis Ei 1 0 Bl 00 Be 00 BS 00 BE 00 BS Go Be 10 BF OD Be Op BS gg Bit 00 hype a ty ree Qemominsice Coethinsents 41 Ob et 1 0 Al 0 Ae gi A gi dt 00 Ai O00 ee 00 Ar OO ae 00 Ag 00 410 O00 E BDA r Sanpa i 00 k d 00 0 OO fo OO Afhi1 Please note that the following notation has been used throughout this document Block names bold and italic e g Plot Drop down menu item name large bold font e g Basic Blocks Button third brackets e g update Option to be chosen by user in a dialog box of a block inverted comma e g Gain Fig 4 Dialog windows in J DSP M1 4 1 3 Example J DSP laboratory assignment This assignment assumes some familiarity with basic DSP concepts It continues from section 1 2 to build on the block diagram created there Start with the block diagram of Fig 4 Let us now form a signal using the signal generator Double click inside the SigGen box and a dialog window will appear This window is shown in Fig 5 If you do not see a dialog window you are using an older Internet browser and must download the newest version of Netscape or Internet Explorer and start over Use Internet Explorer 5 5 or later or Navigator version 4 6 or later with its Java plug in HL Signal Generator Signal Generator Signal Preview Mame al Signal Rectangular f Triangular Delta Gain Random celf Defined a re 10 20 30
11. J DSP A full featured educational tool beeeeeeeeeseeseeseese Existing J DSP Prototype Sesecccccsesccececcceeseeees Fig A5 OF J DSP Editor Seles File View Help Demos Ie a re ae dee 5 EXISTING FUNCTIONS Sm ale cam Speech I Run simulation FFT IFFT PkPking Magn Phase Streaming video Sig Gen Sig Gn J DSP simulation FFT Flot E d E SigGeniLy o 2 D Amplitude scale linear dB ne E an Pd coor A Java Ooplat Window Student notes Lecture notes J DSP HTML lecture interface A a and report in HTML eo a e m ee ONO Plat Orda Y as Fn en v IcrapnivaluessStats Close Help e eeeeeeeeeeneeoenoeeeoeoeneoeoeeee ed See ce ee ses ee se Oe 8 2 ee ee ee eee ie a aa Pe eR Lab 2 concentrates on the Fast Fourier Transform FFT FFT lecture J DSP links E Consider the symmetries in the following signals We want to see how these symmetries affect FFT Related Te rene Related web page content 1 Prof Smith s web page spectra l 2 FFT dedicated web site I H Related books l 1 Classroom text l 2 The FFT transform by E l Ok I a E a a OR i CO ane Nd ae RED EP ae I Planned GUI for Comprehensive Delivery of Lectures Simulations Labs 7 Laboratory specific evaluation questions This is a list of laboratory specific evaluation questions students were required to complete after fin
12. Loizou Speech Recognition Using Minimum Error Classification Speech Communication vol 30 pp 27 36 January 2000 J 14 S Ahmadi and A S Spanias Cepstrum Based Pitch Detection Using a New Statisctical V UV Classification Algorithm IEEE Trans on Speech and Audio Vol 7 No 3 pp 333 338 May 1999 J 15 P Loizou and A S Spanias Improved speech recognition using a subspace projection approach IEEE Trans on Speech and Audio vol 7 no 3 pp 343 345 May 1999 J 16 S Ahmadi and A S Spanias A New Phase Model for Sinusoidal Transform Coding of Speech IEEE Trans on Speech and Audio vol 6 no 5 pp 495 501 Sept 1998 J 17 M Deisher and A S Spanias Speech Enhancement using state bases estimation and sinusoidal modeling Journal of Acoustical Society of America vol 102 2 pp 1141 1148 Aug 1997 J 19 J 20 J 21 J 22 N23 J 24 J 25 J 26 J 27 J 28 J 29 J 30 J 31 J 32 J 33 P Loizou and A Spanias High Performance Alphabet Recognition IEEE Trans on Speech and Audio vol 4 no 6 pp 439 445 Nov 1996 P Loizou and A Spanias Improving discrimination of confusable words using the divergence measure Journal of Acoustical Society of America Vol 101 2 pp 1106 1111 Feb 1997 Q Shen and A Spanias Adaptive Active Sound Reduction Noise Control Engineering Journal J44 6 pp 281 293 Nov 1996 A Spanias
13. Loizou and A S Spanias Context dependent Modeling in Alphabet Recognition IEEE Proceedings of the International Symposium of Circuits and Systems ISCAS 94 pp 189 192 London May 1994 S Karkada C Chakrabarti and A S Spanias High Sample rate Architectures for Block Adaptive Filters IEEE Proceedings International Symposium of Circuits and Systems ISCAS 94 pp 131 134 London May 1994 R Fulchiero and A S Spanias Speech Enhancement Using the Bispectrum IEEE Proc International Conference on Acoustics Speech and Signal Processing ICASSP 93 pp 488 491 Minnesota March 1993 Ines Jebali P Loizou and A S Spanias Speech Processing Using Higher Order Statistics IEEE Proceedings of the International Symposium of Circuits and Systems ISCAS 93 pp 160 163 Chicago May 1993 A Spanias and R Fulchiero Speech Enhancement using the Least Squares Bispectrum Reconstruction Algorithm IEEE Proceedings of the International Conference on DSP and CAES pp 434 441 Nicosia Cyprus 1993 12 C 56 C 57 C 58 C 59 C 60 C 61 C 62 C 63 C 64 C 65 C 66 C 67 C 68 C 69 C 70 K Tsakalis M Deisher A Spanias System Identification Based on Bounded Error Constraints Proceedings of the International Conference on DSP and CAES pp 75 84 Nicosia Cyprus 1993 A Elia C Pattichis W Fincham A Spanias and L Middleton Autoregressive Spectral Modeling of Mo
14. Mountain Bioengineering Symposium Biloxi Mississipi USA April 11 13 2003 in press E7 C 8 C 9 C 10 C l2 C 13 C 14 C 15 C 16 C 17 C 18 C 19 C 20 C 21 Rajeshkumar Venugopal A Prasad K Narayanan A Spanias amp L D Iasemidis Nonlinear noise reduction and predictability of epileptic seizures Proceedings of IASTED International Association of Science and Technology for Development International Conference Palm Springs California USA Feb 24 26 2003 pp 240 245 Shivkumar Sabesan K Narayanan Awadhesh Prasad A Spanias and L D Iasemidis Improved measure of information flow in coupled nonlinear systems Proceedings of LASTED International Association of Science and Technology for Development International Conference Palm Springs California USA Feb 24 26 2003 pp 329 333 Shivkumar Sabesan K Narayanan Awadhesh Prasad A Spanias J C Sackellares amp L D Iasemidis Predictability of epileptic seizures A comparative study using Lyapunov exponent and entropy based measures Proceedings of the 40th Annual Rocky Mountain Bioengineering Symposium Biloxi Mississipi USA April 11 13 2003 in press T Thrasyvoulou K Tsakalis and A Spanias J DSP C A Control Systems Simulation Environment For Distance Learning Labs And Assessment 33rd ASEE IEEE Frontiers in Education Conference Boulder CO November 5 8 2003 V Atti and A Spanias A Simu
15. October 1995 Min Tau Lin A S Spanias and F Tiong Robust Speech Recognition Based on Minimum Error Classification and Weighted Projection Measure Proceedings of International Conf on Signal Processing Applications and technology ICSPAT pp 2006 2009 Boston October 1995 Edward Painter and A S Spanias A Software Tool for Understanding and Evaluating Standardized Speech Coding Algorithms International Conference on Digital Signal Processing DSP 1995 pp 850 857 Limassol June 1995 P Loizou and A S Spanias Improved Speech recognition Using the Weighted Average Divergence Measure International Conference on Digital Signal Processing DSP 1995 pp 90 95 Limassol June 1995 K Kitsios A S Spanias and B Welfert Optimum Block Modified Covariance Algorithms for Spectral Analysis 3rd Mediterranean Symposium on New Directions in Control and Automation pp 398 405 Limassol July 1995 G Nair and A S Spanias Fast Adaptive Algorithms Using Eigenspace Projections 1994 Asilomar Conference pp 1520 1524 Monterey October 1994 M Deisher and A S Spanias Speech Enhancement using a State Based Transform Model 1994 Asilomar Conference pp 1242 1246 Monterey October 1994 K Daroudi and A S Spanias Speech Intelligibility Improvement using the Intel Technique Proceedings of International Conf on Signal Processing Applications and technology pp 114 117 Dallas October 1994 P
16. Portions of this work have integrated in a speech coder developed later for a funded project This project satisfied the requirements of the EEE 490 course Comparative Study of Filter Design Algorithms Student J Snider The project entailed developing and comparing several digital filter design algorithms Graphical demonstrations were also given This project satisfied the requirements of the EEE 490 course Programmable Signal Generator Student Bruce Negley This project entailed a real time hardware implementation of programmable signal generator This project satisfied the requirements of the EEE 490 course Analog Television Standard Conversion A Digital Processing Approach Student Brian Crawford This projects entailed real time hardware implementation of a digital converter PAL to NTSC This project satisfied the requirements of the EEE 490 course Adaptive Signal Processing Algorithms James Chan This project entailed developing software for a real time visual demonstration program on the PC of the adaptation process associated with LMS algorithms The demo program shows in color the adaptation of the poles of a second order adaptive system to the poles of a time varying system This software is being used in some of my classes for demonstration of an adaptive process This was part of the EEE 490 senior project course MATLAB Implementation of the IS 54 speech coding algorithm Student Renos Ioanno
17. Spanias A block time and frequency modified covariance algorithms for spectral analysis IEEE Trans on Signal Processing vol 41 no 11 pp 3138 3153 Nov 1993 A Spanias principal investigator 16 A S Spanias and P Loizou A mixed Fourier Walsh transform scheme for speech coding at 4 kbps Proc IEE Part I Communications Speech and Vision vol 139 5 pp 473 481 Oct 1992 A Spanias principal investigator and Ph D advisor 17 A S Spanias A Hybrid Transform Method for Speech Analysis and Synthesis Signal Processing Vol 24 pp 217 229 Aug 1991 A Spanias principal investigator 18 W Mikhael and A S Spanias A fast frequency domain adaptive algorithm Proc of the IEEE vol 76 no 1 pp 80 82 Jan 1988 Complete Curriculum Vita Andreas Savva Spanias Professor Department of Electrical Engineering Background Personal U S Citizen Married to Photini Spanias Ed D Two Children John 13 and Louis 10 Education e Ph D 1988 Dept of Electrical and Computer Eng West Virginia University e M S E E 1985 Dept of Electrical and Computer Eng West Virginia University e 6B S E E 1983 Dept of Electrical and Computer Eng West Virginia University e HTI Diploma 1979 Diploma in Electrical Engineering Higher Technical Institute HTI Nicosia Cyprus Areas of Teaching and Research Teaching Digital Signal Processing Linear Systems Speech Processing Adaptive Si
18. Spanias is one of three directors on this project Planning Grant for a Center for Collaborative Research in Learning Technologies NSF 50 000 PI A Spanias and 15 other CO PIs Multidisciplinary Research on the Next Generation Multimedia Technologies for Interactive Distributed Learning State of Arizona ASU VPR Multidisciplinary Initiative Committee Pre proposal already approved 150 000 for three years External Funded Equipment Proposals PI A S Spanias For the Development of Speech Coding Algorithms for Teleconferencing Applications Two Intel Pentium Multimedia PCs Amount 12 500 00 Date Feb 1995 PI A S Spanias Various DSP boards donated to A Spanias after proposal by Motorola AT amp T Intel DSP Group boards Totaling Approximately 34 254 00 Dates 1992 1994 20 Internal Funded Research Projects e PI A S Spanias Multimedia Education over the Internet CIEE FGIA May 1996 May 1997 6 000 e PI A S Spanias Speech Processing Based on Higher Order Statistics HOS DWR B708 Agent Faculty Grant In Aid Program ASU January 1991 December 1991 5 000 00 e PI A S Spanias Low Bit Rate Speech Coding DWR B575 Agent Faculty Grant In Aid Program ASU January 1989 December 1989 3 000 00 Internal Awards and Equipment Support Funds e Funds for Equipment Upgrades and Acquisitions in the Speech Processing Lab Amount 32 000 Engineering College EE TRC 1995 96 e Funds for E
19. Step 3 To view the signal in the frequency domain insert an FFT block between the Filter and the Plot boxes as shown below in Fig 7 The FFT block can be found under the Freq Blocks menu Fig 7 Source filter simulation with FFT at the output Step 3 1 Set the Filter parameters and input as per step 2 4 Double click on the FFT block and change the FFT size to 256 points and then press Close Now you can see the magnitude and the phase of the signal in the frequency domain The magnitude has a sharp peak approximately at 0 31 the frequency of our sinusoidal signal 0 1x3 1459 Step 3 2 Change the sinusoidal frequencies as per steps 1 2 and 1 3 but with pulse width 256 What do you observe Step 3 3 Delete the Filter block Set the sinusoidal frequency in SigGen as per step 1 1 but with pulse width 256 Now create a second SigGen block and a Mixer block Your editor window should contain a block diagram that looks like the one in Fig 8 M1 7 Fig 8 Sinewave plus noise simulation Change the name of the first SigGen block to Sinusoi the second SigGen block to noise and the Plot block to SigNo1 The names are restricted to six characters Following that we edit the SigGen block called noise Open the dialog window and change the signal type to random Choose a variance of 4 and extend the pulse width to 256 samples in order to have noise over the ful
20. Volume 3 pp 210 213 2002 Salvatore Bellofiore Jeffrey Foutz Constantine A Balanis and Andreas Spanias Smart Antennas for Wireless Communications 2001 IEEE Antennas and Propagation International Symposium Boston Massachusetts vol 4 pp 26 29 July 2001 T Painter and A S Spanias Perceptual segmentation and component selection in compact sinusoidal representations Proc IEEE International Conference on Acoustic Speech and Signal Processing ICASSP 2001 Salt Lake City May 2001 A S Spanias and Fikre Bizuneh Development of new functions and scripting capabilities in java dsp for easy creation and seamless integration of animated dsp simulations in web courses Proc IEEE International Conference on Acoustic Speech and Signal Processing ICASSP 2001 Salt Lake City May 2001 Jeff Foutz and Andreas Spanias Adaptive Modeling and Control of Smart Antennas pp 859 862 Proc MIC 2001 Innsbruck Feb 19 22 2001 10 C 22 C 23 C 24 C25 C 26 C21 C 28 C 29 C 30 G52 C 33 C 34 C 35 C 36 C 37 C 38 Salvatore Bellofiore Jeff Foutz Israfil Bahceci Constantine A Balanis Andreas Spanias Tolga Duman and James T Aberle Smart Antennas for Mobile Platforms International Union of Radio Science URSI 2001 Boulder Colorado pg 243 Jan 2001 S Ahmadi and A Spanias Minimum Variance Phase Prediction and Frame Interpolation Algorithms for Low Bit Rat
21. are b 0 75 and R 500 M9 4 M9 4 Block name Reverberation Notation Reverb Description This block implements a reverberation effect on the input signal Reverberation is obtained by mixing the input signal with the delayed versions of its feedback The effect of the feedback results in multiple echos Pin assignment rs Dialog window s Feedback Delay Eno Feedback Gain 0 5 Close Update Help Java Applet Window a Reverb dialog window Script use Name Reverb Example code lt param name 3 value B3 Reverb 2 4 gt Equation s Implemented y n x n b y n R R feedback delay in samples b is the attenuation constant Ibl lt 1 M9 5 M9 5 Block name Reverberation Demo Notation Reverb Demo Description This block is a demonstration of the reverberation effect simulating five specific cases given by Cavern delay 600 gain 0 7 Dungeon delay 160 gain 0 8 Garage delay 240 gain 0 4 Acoustic Lab delay 128 gain 0 6 and Closet delay 40 gain 0 1 Pin assignment rsp Dialog window s Cavern Dungeon C Garage C Acoustic Lab Closet Close Update Help Java Applet Window a Reverb Demo dialog window Script use Name ReverbDemo Example code lt param name 4 value B4 ReverbDemo 4 3 gt Equation s Implemented y n x n b y n R R feedback delay in samples b is the attenuation c
22. exp Example code lt param name 3 value B3 exp 3 1 gt Equation s Implemented x n y n e x n input signal y n output signal M4 5 M4 5 Block name Power 10 Notation 1040 Description This block calculates the power 10 of the input signal Pin assignment Input signal Output signal a O e ON d Dialog window s None Script use Name 10pow Example code lt param name 3 value B3 10pow 3 1 gt Equation s Implemented y n 10 x n input signal y n output signal M4 6 M4 6 Block name Sum of squares Notation Square Description This block calculates the sum of squares of the two signals at its inputs The coefficients a and b are user defined Pin assignment Input signal x n zz Dialog window s Square Function a x Sum of squares of two signals 1 amp x2 qs ey Bel ha se fa ea coefficient a he coeficient h ho Close Update Help Java Applet Window a Square dialog window Script use Name square Example code lt param name 3 value B3 square 3 1 gt Equation s Implemented y n ax n bx n x n input signalat pin 1 X2 n input signal at pin 2 y n output signal a b are the weights entered by the user M4 7 Section M3 Basic blocks These blocks appear at the top of the simulation area Table of blocks Block notation Descr
23. lt PARAM gt tags in this program 2 lt param name 0 value BO siggen 3 1 gt 3 lt param name 1 value B1 plot 5 1 gt 4 lt param name 2 value C 0 4 1 0 gt The above program establishes a SigGen block in the 3 1 position and a Plot block in the 5 1 position of the editor frame The last lt PARAM gt tag connects the two blocks by connecting SigGen block s 4 pin and Plot block s 0 pin Opening the html file and starting J DSP will give the flowgram of figure 5 TZ J DSP Editor k Oj x Fie view Help Demos Statistics Window Mixer D Sampiing U Sampling Convolution Sig Gen SigGen L ae Coeff i Junction Filter Freg Resp Plot Plot2 Snd Player Figure 5 A flowgram with connections 3 4 Passing parameters to parts P is used to denote parameter passing in a specific block A lt PARAM gt tag containing P in its value will cause the set of parameters that follow to be passed to the loaded block The number J M10 7 concatenated with P is the number originally given to the part when created with the B command For example in the code below line 3 shows how a signal generator part is created and given the number 0 lt param name 0 value BO siggen 1 3 gt Then line 9 is used to pass parameters to the signal generator part by placing a O next to the P command J Parameters to be passed to the block lt param name 2 value P0
24. lt l START PART PARAMATERS DO Ng lt param name 11 value P0 20 10 0 lt param name 1 2 value P1 0 b Java Applet Wi lt param name 1 3 value P2 c cont 1 4 yalue P3 Java Applet Window aie aeli m a 1 Close 1 Prepare demonstration in J DSP 4 Add your own educational content bw Hep Demos a sy pel Scat ee CMM Filter Blocks PLANNED FUNCTIONS DISCLAI ER rzpuconen Pzero eroen wgbesin kaiseroeson Juve J J DSP SCRIPT lt applet CODE JDs lt param name lt param name 4 va lt END PARTS apa m x 3 DSP tutorial 3 DER ar File Edit View Favorites Tools Ip gt gt FN 7 lt gt Se P Search ST Favorites a Media gA gT Se yE ine A Address wiv myclass asu edul Z transform labo f IEF DER In this lab we use thie EXIS ING FUNCTIONS ECTE Speech 1 gt transform of various PZI lab the Filter block function of the follo num Sig Gen i START P TS SigGent L lt param name 0 ve lt param name 3 ve nS coer Junction PZ Placement Block Filter lt START CONN Freq Resp lt param name 5 va Ey Frequency Response Version 1 0 i Id lt param name 6 va Name a lt param name 7 va Java Applet Window lt I e END CONNEC nude Editor Sng Player E Quantizer a J DSP concept by Prof Andreas S
25. stationary data using a periodogram and a correlogram is assigned The performance characteristics of the two estimators are evaluated in terms of variance and resolution capability This process is repeated for the AR spectral estimator 4 LAB SUBMISSION PROCEDURE All the Lab assignments are accessed on line and students submit their work on the internet Each of the students in the EEE407 DSP class is given a user name and password for a lab account Using lab account students complete an on line media rich report that includes a quiz that covers the laboratory material The submitted electronic report contains responses to multiple choice questions dialog boxes for writing qualitative comments and facilities to upload the graph files and equations in gif format Upon submitting the report all of the student s answers comments and graphs are placed together in a static HTML that corresponds to each student s ID Part of the grading is done automatically while part of it requires instructor intervention The automatic part is processed by a UNIX shell script on the server computer that grades the answers of the multiple choice and true false questions 5 PRE POST LAB ASSESSMENTS In addition to the general assessment of J DSP and the pertinent exercises carried in EEE407 in the Fall of 2002 in the spring of 2003 we started running specific pre and post lab assessments for each lab The purpose of the assessments is to survey a
26. with an idea as to whether the pertinent labs have been useful and helpful to them In pre post lab assessment the questions posed were technical and are set to test the student s level of knowledge before and after performing the J DSP lab In the general assessment the questions 13 are less technical and are posed to get feedback from the students regarding the usefulness of the J DSP software and associated exercises Overall the responses were very promising From Table and Figure 6 we see that 95 of the users appreciated the various features of J DSP as an internet based simulation tool From Figure 6 it is clear that it took most 70 of the users less than half an hour to learn using the software In fact 85 5 of the users agreed that they would consider using J DSP for DSP simulations Lab specific general assessment was done on Labs 1 through 4 in the Fall 02 semester the current semester is spring 2003 TABLE I STATISTICS BASED ON USER EVALUATIONS OF J DSP TOOL gt L g n Q Evaluation questions E Bb D BS ae oD xe BR 17 ZU g BBY 1 Establishing and connecting blocks is 53 39 7 1 0 easy 2 The graphical interface of J DSP is intuitive and user friendly k o 3 i 3 Setting up the required lab simulations 40 a2 8 0 0 was easy From Table 2 we see that most of the students above 90 agreed that the J DSP labs helped them understand the DSP related concepts Also from Table 3 it is evident that more
27. 24 10 0 3 0 0 9 0 0 0 2 a Triangular Yes null gt Summarizing P denotes that parameters are to be passed and 0 gives the part number to which they are to be passed to The set of parameters to be passed to the part then follows The set of parameters passed to each part differs but the form of the parameter passing line is always the same and is given by lt param name x value Py io 1 In o d1 dn So0 1 Sp Do 04 Dn gt where x is the script line number and y is the number given to the part when it was created Here i stands for integers e g 2 d for doubles e g 3 52 s for strings text and b for Booleans true or false Note that this line has to be typed exactly as shown above with the commas and tildes in the exact position even when a certain type of variables is not present notice that the two consecutive tildes at the end of line 9 are there even though no Boolean variables are present For explanatory purposes table 1 describes the parameters of a signal generator SigGen block However since it could become cumbersome to manually pass parameters to a part no other tables have been added and the user is urged to automatically generate scripts that pass parameters to parts This will save both time and debugging efforts The following paragraphs describe how to use table 1 As mentioned earlier the line necessary to pass parameters to a part has the following form lt param name x
28. B Jonsson Dept Electr Eng ASU May 1990 Stefan is with Honeywell in Phoenix Block Adaptive Prediction Algorithms G A Sarrouh Dept Electr Eng ASU May 1990 Until 1993 George was employed in Tempe Time and Frequency domain Adaptive Noise Cancellers V Rhodes Dept Electr Eng ASU May 1990 Val is with EF Data in Tempe 23 Student Dissertations Theses Supervision In Progress Language Modeling for Voice Recognition Ajith Mekkoth Ph D DSP Algorithms for Smart Antennas Jeff Foutz Ph D Array Antennas for CDMA systems Steve Miller Ph D Speech Modeling using microradars Atti Venkataraman Ph D Beamforming Algorithms Ashwin Ph D Digital Communications Ghassan Malouli Ph D Psychoacoustic models in speech coding Yu Song Adaptive Beamforming Algorithms Thrassos Thrassyvoulou M S Adaptive Equalization Algorithms Costas Constantinou M S Communications Algorithms for Java DSP Fikre Bizuneh M S Analysis of MP3 and its use in Noise Reduction Ryan Pintoi M S Analysis of DNA Data using Linear Prediction Niranjan Chakravarti M S Extensions on Java DSP in Communications Atti Venkataraman M S Microphone Beamforming Seth Benton M S Students that have been or are being funded from sponsored research projects 24 Professional and Scientific Service Local IEEE Activities IEEE Communications and S
29. Gaussian Rayleigh Table 1 Signal Generator parameters table The parameter passing line can be skipped if no parameters are to be passed to the part which then loads with its default parameters Finally figure 6 shows a J DSP script for creating a simulation using a filter and a coefficient block lt applet CODE JDsp class width 400 height 250 gt lt param name numCommand value 12 gt lt START PARTS gt lt param name 0 value BO siggen 1 3 gt lt param name 1 value B1 filter 3 3 gt M10 9 lt param name 2 value B2 coeff 3 5 gt lt param name 3 value B3 plot 5 3 gt lt l END PARTS gt lt START CONNECTIONS gt lt param name 4 value C 2 3 1 2 gt lt param name 5 value C 0 4 1 0 gt lt param name 6 value C 1 4 3 0 gt lt END CONNECTIONS gt lt START OPEN DIALOGS gt lt param name 7 value O0 3 gt lt END OPEN DIALOGS gt lt part parameters gt lt param name 8 value PO 20 10 0 1 0 0 9 0 0 0 2 a Rectangular No null gt lt param name 9 value P1 gt lt param name 10 value P2 1 0 1 8 2 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 c gt lt param name 11 value P3 d cont Magn linear false gt lt part parameters gt lt applet gt Figure 6 3 5 Opening Dialog Windows O is used to open the dialog box of a block The effect is
30. In final Preparation Java DSP Software Concept 33 000 lines ISBN 0 9724984 0 0 Copyright A Spanias 2002 Invited Contributions in Books Book Chapters e Andreas Spanias Chapter 3 Speech Coding Standards pp 25 44 Invited Academic Press Ed G Gibson ISBN 2000 0 12 282160 2 e A Spanias Speech Coding for Mobile and Multimedia Applications Invited Chapter in Digital Signal Processing Technologies Critical Technology Reviews CR57 Eds P Papamichalis and R Kerwin pp 115 144 SPIE Press Washington 1995 e A Spanias Speech Coding for Wireless Applications Invited Chapter in Microsystems Technology for Multimedia Applications Eds Sheu et al Chapter 4 pp 257 279 IEEE Press New York 1995 15 Other Publications Scientific Reports Technical Reports for Funded Research Projects R 1 Smart Antennas for Mobile Networks Progress Report Submitted to the National Science Foundation C Balanis A Spanias et al May 2000 R 2 Universal Speech and Audio Coding Using a Sinusoidal Signal Model A Spanias S Ahmadi and T Painter To Intel MRC ASU TRC Technical Report TRC SP ASP 9603 September 1996 R 3 Design Analysis and Implementation of the GSM Channel Codec and Modem A Spanias J Sadowsky K Daroudi A Mekkoth S Azizi To Intel MRC ASU TRC Technical Report TRC SP ASP 9602 August 1996 R 4 A Software Tool for the Evaluation of Speech Enhancement Algorithms on the Intel Pentium Pro
31. Input Malespeaker l hal a re ae Gain Malespeaker l Femalespeaker l Music sine_1 sine_2 Framesize So Overlap 250 je 055 25 C 50 Samples ZE gt Ee SS ea Pause Rerun Amplitude Normalized to I Show Graph i Current frame All frames Help Close a SigGen L dialog window Script use Name siggen L Example code lt param name 3 value B3 siggen L 3 1 gt M2 4 M2 3 Block name Coefficient Notation Coeff Description This block allows the user to enter filter coefficients A maximum of 11 coefficie nts can be used Coefficients can be entered in tabular form or by line form as shown below The by line option provides an easy way to cut and paste coefficients from other sources Pin assignment Description a Coefficients numerator and denominator Dialog window s Filter Settings Filter Settings Filter Setting Filter Setting Name fk Nare Ja bi 1 0 bE 0 F7 bE 00 big al E bo E Select Select al 1 41 b1 0 7 display OG 0 0 IF 8 0 BEBE 0 0 Igy display az 1 0 pz 0 0 l a3 0 0 b3 0 0 ype AM 1 0 aft 1 41 af2 1 0 a ad 0 0 b4 0 0 by line a5 0 0 bS 0 0 al 00 aff 0 0 al 0 0 afd 0 af jo 0 hE oo ar 0 0 bi 0 0 ag 0 0 be 0 0 oo bd 0 0 b0 b10 a Insert iil Jt 0 7 Q L 0 jo 0 610 o Close Update Close Update Help Java Applet Window Java Applet Wi
32. Interface and Navigation There are only a few basic repetitive steps one needs to perform in order to construct simulations with J DSP By pressing a button the user can select and then drag and drop any block where desired Connections can be constructed simply by dragging the mouse from a block s output to another block s input Last but not least all blocks have a dialog window that opens when the user double clicks on it All dialog windows are consistent and include a Help facility 9 Engineering Content 9 1 Accuracy The software has proven to be accurate and quite stable Before releasing it for use a series of tests have been performed to validate the correctness of the code Procedures are in place to report and correct bugs 9 2 Appropriateness J DSP is an educational on line simulation environment developed to support labs Although it is not the traditional multimedia courseware type it falls within the context of software developed for education by educators and we hope that the NEEDS panel will evaluate this novel abware concept for the NEEDS award Bibliography copies of the publications have been included in the same order in Appendix E 1 Clausen A Spanias A Xavier A A Java Signal Analysis Tool for Signal Processing Experiments JEEE International Conference on Acoustics Speech and Signal Processing ICASSP 98 DSP 16 Seattle May 1998 2 Spanias A et al Development of a Web based Signal and Speech
33. J DSP editor environment By connecting blocks together a variety of DSP systems can be simulated Signals at any point of a simulation can be examined and block parameters can be edited through dialog windows TE J DSP Editor me _ lolx File View Help Demos EXISTING FUNCTIONS Tamm Specch FET iret Pk Pring Magn Phase Block set selection SigGeniL Push buttons to select blocks Blocks Junction Filter Freq Resp Flot Error warning Plot messages and Plaver ava Applet Window J Fig 1 J DSP Graphical User Interface System execution is dynamic which means that any change at any point of a system will automatically take effect in all related blocks Any number of block windows can be left open to enable viewing results at more than one point in the system The NEEDS panel reviewer may visit the web site and first view some of the instructional AVI files to become familiar with J DSP A step by step process has been posted on the web http jdsp asu edu to guide the reviewers through J DSP Student users are typically assigned a small signal processing task as part of their introductory lab Appendix D in EEE 407 and then asked to run specific simulations and make measurements to verify theory using J DSP In class evaluations students reported that it typically took them just a few minutes to become familiar with the J DSP GUI A J DSP Quick Reference Getting
34. Nov 1988 W B Mikhael A S Spanias F H Wu Fast Frequency Domain Implementation of a Block IIR Filter with Applications IEEE International Symposium on Circuits and Systems ISCAS 88 Conf Proc ISCAS 88 pp 285 288 Espoo Finland June 1988 A S Spanias and W B Mikhael Implementation of the Optimum Block Adaptive Algorithm in the Frequency Domain IEEE International Symposium on Circuits and Systems ISCAS 87 Conf Proc ISCAS 87 pp 426 429 Philadelphia May 1987 W B Mikhael and A S Spanias Performance Enhancement of the Frequency Domain LMS Adaptive Algorithm IEEE International Symposium on Circuits and Systems ISCAS amp 6 Conf Proc ISCAS 86 Vol 1 pp 349 352 San Jose California May 1986 W B Mikhael A S Spanias F H Wu An Adaptive Pole Zero Predictor IEEE International Symposium on Circuits and Systems ISCAS 85 Conf Proc ISCAS 85 Vol 3 pp 1107 1110 Kyoto Japan June 1985 W B Mikhael A S Spanias and F H Wu ARMA Modeling Using the Linear Predictor International Conference on Acoustics Speech and Signal Processing ICASSP 85 Conf Proc ICASSP 85 Vol 4 pp 1505 1508 Tampa Bay Florida April 1985 K Tsakalis M Deisher A Spanias FIR Adaptive Filtering Based on a Bounded Error Criterion Proc Asilomar Conference on Circuits Systems and Computers Invited pp 15 18 Monterey Nov 1992 A S Spanias and F H Wu Speech Coding and Speech Re
35. Processing Laboratory for Distance Learning ASEE Computers in Education Journal pp 21 26 Vol X No 2 April June 2000 3 Spanias A and Bizuneh F Development of new functions and scripting capabilities in java dsp for easy creation and seamless integration of animated dsp simulations in web courses Proc IEEE International Conference on Acous c Speech and Sign Proc ICASSP 2001 pp 2717 20 Salt Lake City May 2001 4 Thrasyvoulou T Tsakalis K and A Spanias J DSP C A Control Systems Simulation Environment for Distance Learning Labs and Assessment 33rd ASEE TEEE FIE 03 Conf Boulder Nov 5 8 2003 5 Zaman M Papandreou Suppappola A and Spanias A Advanced Concepts in Time Frequency Signal Processing made Simple 33rd ASEE TEEE FIE 03 Boulder Nov 2003 6 Spanias A Ahmed K Papandreou Suppappola A and Zaman M Assessment of the Java DSP J DSP On Line Laboratory Software 33rd ASEE IEEE FIE 03 Boulder Nov 2003 7 Spanias A T Thrassyvoulou C Panayiotou Y Song Using J DSP to Introduce Communications and Multimedia Technologies to High Schools 33rd ASEE IEEE FIE 03 Boulder November 2003 8 Yasin M Karam L and Spanias A On Line Laboratories For Image And Two Dimensional Signal Processing 33rd ASEE IEEE FIE 03 Boulder Nov 2003 9 Atti V and Spanias A On line Simulation Modules for Teaching Speech and Audio Compression 33rd ASEE IEEE FIE 03 Boulder Nov 2003
36. Started Guide was also disseminated 2 Target Audiences e Principal Audience The J DSP was originally intended to provide on line computer laboratory experiences to Electrical Engineering students taking introductory DSP classes These are typically undergraduate students at the junior or senior level The version of J DSP that has been completed and submitted for evaluation by the NEEDS committee is the one currently used in the EEE 407 DSP class at Arizona State University This course is a senior undergraduate elective which is also attended by first semester graduate students usually specializing in DSP or Communications The J DSP labs Appendix D that are performed using the J DSP software comprise the one credit laboratory portion of the 4 credit EEE 407 course All the students in that class are required to attempt the J DSP laboratory exercises prepare lab reports and submit them electronically using the tools on the EEE 407 laboratory web site e ASU Colleagues and Students in other Areas With the new functionality in communications 8 controls 4 and image processing 10 my colleagues working in these areas have used some of the J DSP functions and supporting software in assigned homework in their classes Descriptions of exercises and preliminary assessment results are reported in the corresponding papers 8 4 10 e Student Visitors and Practitioners Our J DSP software is open to all students and practitioners worldwide Our w
37. a real world application Such projects include models for compression algorithms used in cellular phones MP3 and JPEG Our class evaluations reveal that J DSP is responsible for building the necessary programming and algorithmic intuition required to perform DSP algorithm projects By the very nature of the J DSP environment and the assigned open ended lab exercises students develop the ability to explore multiple solutions to design problems and extend ideas to other application areas We also have evidence that many of the 407 students attend graduate school and engage in algorithm development in their M S theses 7 3 1 Cognition Conceptual Changes Our DSP class and particularly our pre and post assessment Appendix C reveal that J DSP enhances learning of the key topics covered in the DSP class Statistics and details for each lab are given in Appendix C In the J DSP laboratory students submit electronic reports Portions of the report are graded automatically and students get instant feedback 7 4 The content structure The content in the J DSP lab exercises Appendix D is structured in a manner that follows the progression of topics in any undergraduate DSP course The next laboratory exercise builds on knowledge gained by the previous one The software functions blocks in J DSP are also organized in an intuitive manner Frequently used functions always appear to the left filter Sig Gen etc The rest of the functions are grouped in
38. and Computer Engineering Georgia Institute of Technology School of Electrical and Computer Engineering Gcorgia Institute of Technology Atlanta Georgia 30332 0250 U S A J DSP project at a glance J DSP Editor J DSP A new paradigm in education J DSP is an on line object oriented graphical DSP editor written as a Java applet It creates a new paradigm for education It can be used to simulate DSP systems Users can view the results at any point of the simulation graphically or numerically It provides a simple and user friendly interface TE J DSP Editor File View Help Demos EXISTING FUNCTIONS FFT IFFT PkPking Magn Phase N TT Buttons to select blocks E Sig Gen 2 V Junction Blocks PLANNED FUNCTIONS Speech 1 Z Filter Freg Resp Plot Working area Dialog windows 2DFrequencyResponse A Spectrogram block BlockName a J DSP Existing functionality J DSP Supports HMM Training E E 1 Number of States in HMM 4 v a Name b 2 Select Files for Training ADD Files Reset Fileselection Image Preview Files selected for HMM Training n For Digit 0 Digit 0Gaussian_noise05 wav Digit 0White_nois For Digit 1 gt Digit 1 Gaussian_noise05 wav Digit 1White_nois For Digit 2 gt Digit 2Gaussian_noise05 wav Digit 2White_nois z Adapt Filteri i m 2 teration Humber 2 i 0 959 0 039 0 0 0 0 0 96 0 071 Spectral Estima
39. been freely disseminated to more than thirty institutions worldwide The dissemination process is still ongoing and therefore we can grant NEEDS the permission to non exclusively disseminate the J DSP executable Version 1 Copyright 1997 2003 as packaged on the CD ROM ISBN 0 9724984 0 0 that accompanies this submission packet To evaluate the software please visit http jdsp asu edu and follow the designated link for the NEEDS award Because J DSP is not an e book or a multimedia presentation but a comprehensive online environment we provide a step by step procedure for the reviewers to familiarize themselves and evaluate the labware and all associated materials We will be obliged if the evaluation committee evaluates this software on its intended purpose which is to provide a DSP simulation environment and online laboratory experiences for DSP courses Please note that our labware is not intended to compete with commercial packages such as the world renowned MATLAB Simulink which we also use and promote in our classes at ASU Our freely accessible non for profit J DSP software simply provides an environment only for compact educational simulations on any platform that is accessible through any browser such as the MS Explorer The software was specifically developed for DSP courses Thank you for considering our Engineering labware J DSP Sincerely huvdrecess Spomics Andreas Spanias Ph D Professor and PI J DSP project IEEE Fellow MAIN C
40. been particularly impressed with the tools and exercises and some are using it for routine design and other compact DSP simulations Here are few useful URLs that can be explored by the interested reader For accessing all the lab assignments of DSP course use http www eas asu edu eee407 The J DSP editor can be accessed at http jdsp asu edu The feedback from the users 1s also available by following the links on the J DSP web site REFERENCES 1 Spanias A et al Development and Evaluation of a Web Based Signal and Speech Processing Laboratory for Distance Learning ASEE Computers in Education Journal Vol X No 2 April June 2000 pp 21 26 2 Spanias A et al On line laboratories for speech and image processing and for communication systems using J DSP 2nd DSP Education workshop Pine Mountain GA Oct 13 16 2002 18 Comments on J DSP by EEE407 students obtained from course and lab evaluations The following is a selection of comments or multiple choice answers made by students who used J DSP to perform instructor assigned laboratory exercises It also includes some answers by J DSP users from the industry and academia B 1 Responses of students to multiple choice questions posted in J DSP online assessment Q How long did it take to get familiar with the basics of the J DSP environment A 15 minutes or less 62 Half an hour 53 An hour 32 More than an hour 15 Q Did the demos hel
41. enables on line interactive DSP laboratories Along with the software we have developed several J DSP laboratory exercises that have been posted on the internet Assessment of the EEE 407 labs was carried both on the web and as part of the instructor and class evaluation The web based assessments have been organized into general software assessments general laboratory assessments concept specific lab by lab assessments and differential pre post assessment for each lab Statistical and qualitative evaluations have been compiled for all the J DSP laboratories and are described in the rest of the report Index Terms Assessment of J DSP On line labs filter design ASU EEE407 DSP Course 1 INTRODUCTION Java DSP J DSP is an NSF funded on line software environment that was developed to provide on line laboratory experiences to distance learning and on campus students J DSP consists of object oriented Java software that resides on the internet and enables students to build simple and complex simulations of DSP algorithms The core software environment was initially developed in the late 1990s 1 Since then the J DSP concept has been continuously developed and updated with a series of new functions and on line laboratories 2 as well as other modular web content The software resides on several servers and is used by students taking the DSP class at ASU and in other universities where beta sites have been established Through the years the s
42. enhance student learning in the DSP class Several hands on J DSP computer exercises have been carefully developed not only to reinforce the DSP concepts covered in class but also expose students to complimentary material that 1s not covered in detail neither in class nor in text books The computational nature of these laboratories and the inclusion of real life signals made J DSP particularly useful in providing engineering intuition and valuable hands on experiences On the other hand simulations of quantization effects and manipulations of truncated signals made students aware of the limitations associated with processing real life signals with DSP algorithms Currently EEE407 includes six computer lab assignments that are assigned on a weekly basis 1 Difference equations and the Z Transform 2 Pole Zero Plots and Frequency Responses 3 FIR and IIR Filter Design 4 The Fast Fourier Transform FFT 5 Multi rate Signal Processing and QMF banks 6 Introduction to Random Signal Processing All these labs require on line access to the J DSP editor Each laboratory contains several problems and exercises The students have to complete an on line quiz and submit graphs and comments in the form of an electronic report 3 DESCRIPTION OF THE LABS Lab 1 Difference Equations and the Z Transform The objective of this lab is to introduce the students to the concepts of linear time invariant LTI systems z transforms and the impulse resp
43. filter LEARNING The student learns how an FIR linear phase system behaves in the time domain z domain and frequency domain Such filters are used in Digital Hi Fi audio systems e J DSP Problem 2 5 The student is required to design and simulate a filter with a prescribed pole zero locations LEARNING The student learns inter relationships of z plane and Fourier transforms e J DSP Problem 2 6 The student is required to simulate cascade and parallel configurations of digital filters LEARNING The student learns the advantages of cascade and parallel configurations and associates these schemes with the theory of linear systems A complete copy of the above laboratory exercise is in Appendix D 1 Assessment of this J DSP exercise is presented in Appendix C 1 and in reference 6 4 2 J DSP Software User Manual J DSP is a web based simulation environment that can be used for several purposes To assist students and other casual users to use J DSP we have prepared several documents included in Appendix F 1 e e acomplete manual where each function or block is described Appendix F e An introductory section that explains how to use drag n drop process to establish simulations e A section that shows how to use J DSP scripts to integrate J DSP simulations in HTML content e A Quick Reference Getting Started Guide included in the dissemination package e Several AVI files that have been posted on the web site to show how the s
44. high pass filters are designed based on the pole zero PZ placement method Figure 2 shows the design of a low pass filter using PZ placement In Problem 4 the pole zero locations and the frequency response for an all pass filter 1s examined Lab 3 FIR and IIR Filter Design This exercise examines the four types of symmetric impulse responses that result in linear phase In addition the constraints on the zeros of linear phase filters are studied FIR filter design using the Fourier series and tapered windows are covered Seven problems are assigned in this lab Problem 1 involves the design of FIR filters 1 e the students are asked to observe the frequency response Z domain symmetry and the group delay for these filters Problem 2 deals with the design of low pass filters by truncating the ideal impulse response using windows The following window types are supported in J DSP rectangular Bartlett Hamming Hanning and Kaiser PT Placermen Block Add poleetneics Gisphical gt E L Lo Clore Upehete Hale Shov Coal Wamng Aol et indo lidar ning Applet Wired ov FIGURE 2 DESIGN OF LOW PASS FILTER USING PZ PLACMENT BLOCK IN LAB 2 PROB 3 In problem 3 the students are asked to design high pass filters using the Kaiser window Problem 4 deals with FIR filter design using frequency sampling method In problem 5 the students are asked to design optimal FIR filters using the Parks McClellan algorithm In prob
45. his team developed several functions that not only render this software as a simulation tool but also as an instruction tool The functionality developed includes e J DSP scripting that enables instructors to built J ava simulations for DSP and embed them in the web content eo 3 fs 3 e An interface to MATLAB that enables instructors to port results from MATLAB simulations and embed them in J DSP simulations and on the web _ FROM Panasonic FAX SYSTEM PHONE NO l Jun 7 2003 25 17PM P4 e Demonstration functionality that enables instructors to built demos of DSP algorithms and include them in web pages without engaging into tedious Java programming E oe The NSF project funded in the year 2000 enabled the ASU team to expand the fimctionality of this software to cover other systems courses such as a Analog and Digital Communications systems Statistical signal processing Speech processing Image processing and 2 D DSP Controls systems On a personal note I have been impressed by several features of J DSP and the fact that this software that is essentially an entire programming system with an interpreter was developed by faculty and students with minimal resources I find the J DSP graphical interface to be user friendly and intuitive The concept of building a simulation and establishing signal flow using block diagrams by using a browser is easy to learn and the programming concept is consistent with the way we view sig
46. paper prize award from the IEEE board of directors Andreas Spanias has developed an extensive suite of MATLAB speech coding software tools for public domain use that have been posted on the web and downloaded by more than 5000 students scientists faculty and engineers in the speech processing community The industry collaboration of Andreas Spanias on chip design 3 has had a lasting impact The unique multi processor architecture Intel Phoenix that was designed in part by Andreas Spanias and his ASU team includes two 60172 DSP cores and a micro controller along with a hardware Viterbi accelerator An enhanced version of this architecture is currently being promoted for third and fourth generation wireless cellular systems As part of this work Andreas Spanias and his team have also developed a series of modified linear prediction analysis by synthesis algorithms accompanied by a large body of fixed point software for coding applications An algorithm and portions of this software found their way to the Intel ProShare videoconferencing product As evidence of the great impact of his work for Intel Andreas Spanias received a series of awards from Intel Corporation citing technical leadership and outstanding contributions Registered by Intel Corp Other individual contributions are e Significant contributions in adaptive signal processing time varying spectral estimation 15 adaptive antenna arrays 4 frequency domain adaptive fi
47. plots Q1 The contents of this exercise improved your understanding of the concepts of pole zero plots and frequency responses L Strongly Agree Agree Neutral Disagree OOo Strongly Disagree Q2 You now understand more clearly the relationship of the location of poles and zeros with the frequency response of a system L Strongly Agree Agree Neutral Disagree nann Strongly Disagree Q3 What is the difference between PZ Placement and PZ Plot block Q4 Did you add the poles or and zeros graphically or manually in PZ Placement block Which one do you prefer Why El Q5 Did you experience any problems in terms of connection time to implement etc Please describe El Q6 Was there enough information in the help screens to assist you in using the blocks s Yes s No La Did not use the help screens Q7 Is there any particular concept which you could not grasp earlier from your text book class but became clear from this exercise If so please describe E Q8 Performing the exercises you are now more comfortable with these topics s Yes s No Q9 Setting up the required lab simulations was pretty easy L Strongly Agree Agree Neutral Disagree MO n Strongly Disagree Q10 Can you suggest an exercise along the lines of this one El Q11 Please suggest possible improvements relative to this lab such as redesigning of dialog box of a block what to add or delete etc L
48. responses are equal A 10 99 improvement is noticed here In Question 4 students are asked about the filter type i e low pass high pass etc of a given transfer function We see that only 28 26 of students answered correctly in pre lab assessment and it is indeed poor but the percentage has increased by 44 71 in the post lab assessment In Question 5 students are given four choices regarding the change of magnitude response with displacement of the zeros with respect to the unit circle A noticeable improvement of 25 62 in the post lab quiz is evident m Pre assessment Assessment of Lab2 m Post assessment o Improvement D 1005 5 a 90 SS z gt 80 N Ji x 5080 70 o 2 dD z 60 C wn 3 Se 0 SL 40 522 30 a N c 20 10 0 Q1 Q2 Q3 Q4 Q5 Question Number Fig 2 Sample Results on Pre and Post Assessment of J DSP Laboratory 2 for more results on all J DSP labs see Appendix C 1 5 1 General Remarks on Assessment Continuous feedback helped correct and improve the J DSP software and exercises The students in general found the J DSP concept very convenient and easy to use Concept specific assessments and pre post assessment revealed that several J DSP functions have been particularly useful in communicating key DSP concepts J DSP was proven to be particularly useful in learning issues related with filter design and interpretation of frequency spectra The J DSP visualizations involving pole
49. signal processing simulation tool The purpose of this educational tool which has been originally presented at ICASSP 97 in Munich Germany is to enable distance learners to perform on line Jaboratories on any platform and from any location J DSP in fact established a new paradigm for distance education where computer laboratories and DSP algorithm simulations are performed on the web using a simple web browser J DSP simulations are established using a J ava based object oriented environment where blocks represent DSP algorithms and when blocks are connected signal flow is established and simulations with multiple algorithms execute _ a This educational tool is being used to facilitate the laboratory portion of the DSP _ course at Arizona State for the last six years This course has a distance learning _ audience and the presence of a web based laboratory became very popular among the students particularly those at remote locations A series of laboratories have _ been developed by Prof Spanias as a companion to this exemplary software tool Electronic web based submission software consisting of seviets was also developed at ASU to make the lab report submission and grading web based T this date the students at ASU use J DSP and the web based submission system in their DSP class The original J DSP concept and the submission system have been described by Prof Spanias and his team in the ASEE Computers in Education Journal pp 21 26 Vo
50. terms of filter frequency domain statistical etc see manual in Appendix F The number of functions available in J DSP exceeds those required in the labs and several scripted J DSP demos are available for the student to explore additional topics not only in DSP but also in other areas Explanations on these J DSP demos are available on the web site Students are also encouraged to verify theory using J DSP and several J DSP simulations are run in class to demonstrate advanced topics 7 5 Multimedia and J DSP J DSP offers a variety of visual and audio tools to enhance learning Student feedback has shown that users of J DSP achieved learning using J DSP animations and the J DSP sound player The sound player is used in many cases to reinforce concepts in filtering and signal compression For example we provide functions that connect music to signals and spectra and we promote these J DSP functions for early exposition to DSP in high schools see paper 7 The very nature of Java which is the backbone of J DSP allows for integrated animations in web and streaming content Functions to integrate more and more multimedia are continuously being developed 7 6 Instructional Use Engineering students can consult a detailed J DSP manual that is posted on the internet Every function has explanations and an integrated help screen The J DSP support web site has integrated assessment tools a way to report problems frequently asked questions site and
51. the magnitude of a signal Pin assignment Input signalx n Output signal y n ek re 3 ee ON Dialog window s None Script use Name magn Example code lt param name 3 value B3 magn 3 1 gt Equation s Implemented y n x n F x n input signal y n Magnitude of the input signal M5 5 M5 5 Block name Phase Notation Phase Description This block calculates the phase of the input signal Pin assignment 1 Input signal 2 Phase of input signal pe C 1 2 4 5 Dialog window s Script use Name phase None Example code lt param name 3 value B3 phase 3 1 gt Equation s Implemented x n input signal n phase of the input signal n Zx n M5 6 Section M4 Arithmetic blocks These blocks appear at the top of the simulation area Table of blocks Block notation Mult log10 In exp 10 Square Description Calculates the product of two signals Calculates the Log base 10 of the input signal Calculates the natural log of the input signal Calculates the exponential of the input signal Calculates the 10th power of the input signal Calculates the sum of squares of the two input signals M4 1 Block name Multiplier Notation Mult Description Multiplies the two signals at its inputs Pin assignment Input signalx n Input signal x2 n Outp
52. to explain some of the functions of J DSP is to work through a simple example To start J DSP go to the link http jdsp asu edu click on Start J DSP and press Start in the subsequent dialog window Then if you agree press the I Accept button at the disclaimer window A relatively large Java applet will start downloading to the browser Even though the J DSP working area appears almost immediately it may take 30 seconds or more to establish the first block for a telephone based 28 8 Internet connection However once the first block is established the program should run reasonably fast Adjust the size of the J DSP editor window so that you are able to view the entire work area Press the SigGen button on the left part of the window Move the mouse to the center of the window and click the left mouse button Note that you have created the signal generator block There are two signal generators SigGen for processing a single frame of the signal and SigGen L for frame by frame processing that is typically used in speech processing simulations Similarly create a Filter and a Plot block as shown in Fig 3 Note that blocks cannot be placed on top of one another There are two plot blocks i e Plot single plot and Plot2 two plots For now use Plot Sig Gen SigGen Loew Coeff Fiter Filter lt FreqResp Pit lt Plot ond Player Cluantizer Fig 3 J DSP buttons for a source filter simulation
53. value Py io 1 In do 04 dn So0 1 Sn Do b4 D gt M10 8 All the information that is to replace the i d s and b is given in table 1 specifically for the signal generator block The table is divided into four rows for the variable name type position in the code and the allowed range The variable name is a short description of the variable and is not used in the code The type describes whether the variable is an integer a double a string or a Boolean The position is the exact location where the variable should be placed in the code It is given by a combination of a letter and a number subscripted after it The letter can be i d s or b for integer double string and Boolean respectively The integer denotes the position starting with 0 The range is the allowed assortment of values the variable is allowed to take All variable names with a and a number in parenthesis depend on a particular selection of signal type but nonetheless should be used By replacing the necessary variables in the above generic script line you can get the necessary code to pass parameters to a part Variable Pulse Width Period Time shift Gain Exponential amplitude base 1 Position fio ti fig do 1 256 Mean 2 Frequency 3 Signal type Range gt 0 0 gt 0 0 A 5 digit Rectangular alphanumeric Triangular word Delta Exponential 1 Sinusoid 3 Sinc Random 2 Self Defined Distribution 2 Range Null Uniform
54. you are not able to use Ctrl C to copy please install the latest java virtual machine from Sun M10 2 found at www java sun getjava Instructions are provided in the troubleshooting section of our web site located at http jdsp asu edu Step 4 If you wish to place the applet in an existing HTML file simply paste the script code at the location where you desire the J DSP Editor applet to appear If you do not have an HTML file copy the entire HTML code into a text editor and save the file with an html or htm extension Steps 4 1 and 4 2 provide some extra details and can be skipped if you feel comfortable with what was just mentioned Step 4 1 Saving into an existing HTML file a Use an HTML editor or any text editor capable of reading ASCII files to load the filename html or filename htm file you wish to use for the script If your HTML editor is a what you see is what you get WYSIWYG editor make sure to select to edit the HTML code itself b Place the cursor where desired and paste the applet code using Ctrl V or right click and Paste as shown in figure 3 c Save the file and then load it in a browser Note Do not attempt to view the HTML file through the HTML editor s internal browser as it might not be a fully functional browser and therefore not capable of displaying active content like Java applets K Microsoft FrontPage F My Documents JDSP HTML html _july 17 JDSP start _jdsp htmi q l 3 Oj x File
55. zero diagrams have shown prominent differences in pre and post assessments that lead us to believe that we need to integrate even more animation and develop demonstrations that are dynamic In all EEE 407 students asserted that they have benefited from J DSP and they particularly appreciated the fact that the tool was available on the web from any location Industry students taking the course from remote sites have been particularly impressed with the tools and exercises and some are using it for routine design and other compact DSP simulations 6 Endorsements 6 1 Endorsements from Colleagues Several colleagues have obtained and endorsed J DSP software and recommendation letters are attached at the end of this document Comments from recommendation letters are itemized below e Professor John Proakis who is one of the most successful text book authors in communications and signal processing writes J have been impressed by several features of J DSP and the fact that this software that is essentially an entire programming system with an interpreter was developed by faculty and students with minimal resources e Professor Paul Hasler from Georgia Tech writes J DSP is indeed a unique software tool that will impact many of the courses related to linear systems and signal processing e Professor C L Max Nikias Dean of the School of Engineering at USC writes This software essentially enables DSP courseware with embedded on lin
56. 0 all 100 Unbiased Biased Lags 0 0019414 8 406E 4 4 954E 4 0 0016213 0 0021258 0 001808 8 061E 4 fd Close Warning Applet window Values Close Update Heb Warnina Applet Window a Autocorrelation dialog window and output values Script use Name autocorr Example code lt param name 3 value B3 autocorr 3 1 gt Equation s Implemented N m l ram X x n m x n where m is the number of lags 0 m N If L N a biased autocorrelation sequence is obtained If L N m an unbiased autocorrelation sequence is obtained M7 2 M7 2 Block name Linear prediction coefficients Notation LPC Description This block computes the linear predic tor coefficients LPC based on the Levinson Durbin algorithm Pin assignment Time domain signal x n Autocorrelation sequence r m LP coefficients a Residual signal e n Dialog window s LPC Coefficients Order i 0 LPC coefficients Sm 1 0 AN 1 6936 Aj 1 5616 A 0 7536 ARM 0 0424 AB 0 5339 Al 0 3005 Af 0 0687 AR 0 1798 AS 0 2908 AmO 0 1715 Close Update Help Warning Applet Window a LPC dialog window Script use Name LPC Example code lt param name 3 value B3 LPC 3 1 gt Equation s Implemented p Residual signal is given by e n x n Xa x n 1 i LP synthesis filter is given by H z lj az i 1 M7 3
57. 0 20 Time Shift jo _ Samples Close Update Help Java Applet Window a SigGen dialog window Script use Name siggen Example code lt param name 3 value B3 siggen 3 1 gt M2 3 M2 2 Block name Long signal generator Notation SigGen L Description This block produces 6 types of signals 1 e male speech female speech music white noise colored noise and sinusoid with a maximum data length of 8192 samples The sinusoid option can generate a sum of two sinusoids based on the specified frequencies and amplitudes I desired an option is provided to synchronize the part s two independent outputs The option frame size represents the number of samples n each frame The option overlap allows frames to overlap Possible overlapping schemes are 0 25 and 50 The output plot may be displayed with the signal normalized either with respect to the maximum magnitude of the current frame or the maximum of the entire signal When the colored noise signal is selected a new window is created where filter coefficients that convert white noise to colored noise can be entered The frames can be directed to the output individually gt gt or all together automatic ally gt gt Pin assignment Pi Time domain signal 1 in frames Time domain signal 2 in frames za Dialog window s Long Signal Generator Input Signal Selectthe pin PIN 1 Signal Preview
58. 002 IEEE Donald G Fink paper prize award He also received the 2003 teaching award from the Phoenix chapter for his contributions in on line laboratories using J DSP and has been recipient of various other research awards from Intel Corporation He is Distinguished Lecturer of the IEEE SPS for 2004 and was recently elected Fellow of the IEEE Research Contributions Explained Andreas Spanias has contributed to DSP and particularly to speech processing and related industry applications He developed a unique mixed basis signal analysis synthesis method 17 along with a novel algorithm that selects iteratively the constituent narrowband and broadband basis functions He applied this method to low rate speech coding in 16 and demonstrated improved speech quality relative to other transform coders In another contribution he and his students improved significantly upon the well known sinusoidal speech analysis synthesis model by using non minimum phase modeling techniques 9 along with improved pitch estimation algorithms 7 This work resulted in speech coding implementations that are reported in 5 He also made important contributions to speech enhancement and speech recognition He and his students developed a novel speech enhancement method that uses a harmonic sinusoidal model whose parameters are estimated using a hidden Markov model HMM 10 This perceptually motivated noise reduction method improves significantly upon the state of the art by
59. 2 C1 09 fa C1 fd Q9 If you are a student or faculty please give the name of your university Q10 Which continent are you logging in from L North America L South America La Europe L Asia L Australia L Africa Q11 In your opinion is this type of on line lab concept beneficial for distance learning L Yes L No Q12 Did you experience any problems while starting the J DSP editor s Yes L No Q13 Any comments E Part 2 of 3 Q1 The J DSP menus and blocks are organized in an intuitive and user friendly manner L L Strongly Agree Agree s Neutral E S Disagree Strongly Disagree Q2 Changing the options in the blocks is easy and convenient L Strongly Agree L Agree L Neutral La Disagree L Strongly Disagree Q3 The graphical interface of J DSP is intuitive and user friendly L Strongly Agree Agree Disagree E s Neutral E E Strongly Disagree Q4 Did you find it easy to define a signal using the Sig Gen block Do you suggest any other options jd Q5 Were the options in the Plot block sufficient to clearly visualize the output s Yes s No Q6 When the Plot block was used for a frequency domain graph were you able to zoom and identify spectral peaks s Yes s No Q7 Are the help screens adequate s Yes s No Q8 Would you consider using J DSP for small simulations apart from the lab exercises s Yes s No Q9 Do you think with the help of a simple ma
60. 33rd ASEE IEEE FTE 03 Boulder November 2003 3 e Using J DSP to Introduce Communications and Multimedia Technologies to High Schools 33rd ASEE IEEE FIE 03 Boulder November 2003 On Line Laboratories For Image And Two Dimensional Signal Processing 33rd ASEE IEEE FYE 03 Boulder November 2003 e On line Simulation Modules for Teaching Speech and Audio Compression 33rd ASEE IEEE FIE 03 Boulder November 2003 a toe i applend the efforts of Andon Spanias and his a Panasonic FAX SYSTEM PHONE NO E ee Jun 7 2003 05 18PM PS a f on Modales pee streaming lectures are being ali aie a Mae students to these topics E E benne a and aie waty of the NEEDS a a 7 ee oa a re ohn G Proakis TE EEE Fellow as a sen Foer Chale ait TSS P No E i UNIVERSITY OF SOUTHERN CALIFORNIA ee School of Engineering Office of the Dean C L Max Nikias Zohrab A Kaprielian Dean s Chair in Engineering June 9 2003 The NEEDS Award Committee This is written in support of the J DSP software that is being considered for tbe _ NEEDS award The Java DSP J DSP software has been developed from the University of Southern California Los Angeles Calitornia 90089 1450 Tel 213 740 0617 Fax 213 740 8493 e mail engrdean usc edu web page hitn Awww usc edu deot enaineerina 03p d 800 200 d ground up for education at Arizona State University and is perhaps the first on line
61. 5 Aug 1996 PI Chaitali Chakrabarti CO PI A S Spanias Special Purpose Architectures for Speech Coding Algorithms Phase 2 Sponsor Motorola Inc 15 000 Aug 16 1995 Aug 14 1996 PI A S Spanias Analysis and Implementation of Modem Algorithms on Intel DSP Architectures Sponsor Intel Corp Amount 56 939 00 Aug 1994 Feb 1995 PI A S Spanias Speech Enhancement Algorithms for Mobile Communications Sponsor Intel Corp Amount 37 940 00 DWT 4460 Aug 1994 Aug 1995 PI A S Spanias CO PI C Chakrabarti Speech Coding Algorithms for Multimedia Applications Sponsor Intel Corporation Amount 54 728 Sept 1993 Aug 1994 PI Chaitali Chakrabarti CO PI A S Spanias Special Purpose Architectures for Speech Coding Algorithms Sponsor Motorola Inc 15 745 00 May 16 1994 May 14 1995 PI A S Spanias Image Processing Algorithms for Teleconferencing and Multimedia Applications Sponsor Motorola Inc Amount 45 000 Feb 1 1994 Jan 31 1995 PI A S Spanias Development of Speech Encoding and Recognition Algorithms for the Phoenix Architecture Phase 2 Sponsor Intel Corp Amount 200 229 00 Aug 1993 Aug 1994 19 14 15 16 17 18 19 20 21 22 23 24 PI A S Spanias Speech Enhancement Algorithms for Mobile Communications Sponsor Intel Corp Amount 36 130 00 Aug 1993 Aug 1994 PI A S Spanias Development of Speech Encoding Recognition and Data E
62. 50 No 5 May 2002 J 6 S Bellofiore C Balanis J Foutz and A S Spanias Smart Antennas Systems for Mobile Communications Networks Part 1 Overview of the Antenna Design EEE Antennas and Propagation Magazine pp 145 154 Vol 44 No 3 June 2002 J 7 A Kitsios A Spanias and B Welfert Fast modified covariance algorithm with individual step sizes Signal Processing 82 5 pp 715 7120 June 2002 J 8 S Bellofiore C Balanis J Foutz and A S Spanias Smart Antennas Systems for Mobile Communications Networks Part 2 Algorithms IEEE Antennas and Propagation Magazine pp 106 114 Vol 44 No 4 August 2002 J 9 A Kitsios and A Spanias Fast modified covariance algorithm with individual step sizes Signal Processing Accepted and will appear in 2002 J 10 S Ahmadi and A Spanias Algorithms for Low bit rate sinusoidal coding Speech Communications 34 2001 pp 369 390 June 2001 J 11 Ted Painter and Andreas S Spanias Perceptual Coding of Digital Audio Proceedings of the IEEE pp 451 513 Vol 88 No 4 April 2000 winner of 2002 IEEE Donald G Fink Prize Paper Award J 12 A Spanias S Urban A Constantinou M Tampi X Zhang M Tampi C Stilianou Development of a Web based Signal and Speech Processing Laboratory for Distance Learning ASEE Computers in Education Journal pp 21 26 Vol X No 2 April June 2000 J 13 Min Tau Lin A S Spanias and P
63. A Block Time and Frequency Modified Covariance Algorithms for Spectral Analysis vol 41 No 11 pp 3138 3153 Nov 1993 A Spanias P Loizou G Lim Y Chen G Hu Analysis Synthesis of Speech using the Short Time Fourier Transform and a Time varying ARMA process Trans of the Institute of Electronics Information and Communication Engineers IEICE Vol E76 A No 4 pp 645 652 April 1993 A S Spanias and P C Loizou A Mixed Fourier Walsh Transform Scheme for Speech Coding at 4 KBPS Proc IEE Part I Communications Speech and Vision Vol 139 5 pp 473 481 Oct 1992 A S Spanias A Frequency Selective Adaptive Algorithm Journal of Computers EE Special Issue on Adaptive Signal Processing Editors D Etter and M Ahmadi pp 301 313 Vol 18 No 3 4 May July 1992 A S Spanias S B Jonsson and S D Stearns Transform Methods for Seismic Data Compression IEEE Trans on Geoscience and Remote Sensing Vol GARS 29 No 3 pp 407 417 May 1991 A S Spanias and F H Wu Speech Coding and Recognition A Review Trans IEICE Special Issue on Fundamentals of Next Generation Human Interface Ed Sadaoki Furui pp 132 148 Feb 1992 Names of student co authors that have been supervised by A Spanias J 34 J 35 J 36 J 37 J 38 J 39 J 40 J 41 A S Spanias A Hybrid Transform Method for Speech Analysis and Synthesis Signal Processing Vol 24 pp 217 229 Aug 1991
64. AMPUS COLLEGE OF ENGINEERING AND APPLIED SCIENCES Department of Electrical Engineering PO Box 875706 TEMPE AZ 85287 5706 480 965 3424 Fax 480 965 3837 Java DSP J DSP An On line DSP Simulation Tool enabling Web based Computer Labs J DSP Concept and Software by Prof Andreas Spanias Ph D Multidisciplinary Distance Learning Initiative MIDL Department of Electrical Engineering Arizona State University Tempe AZ 85287 5706 Email spanias asu edu 1 Background Java DSP J DSP http jdsp asu edu is an educational program that enables on line simulations and web based computer laboratories for use in Digital Signal Processing DSP courses The initial version of J DSP has been developed in the ASU MIDL lab and tested in a senior level Electrical Engineering DSP course EEE 407 during the academic year 1996 97 Portions of the software have been displayed in the 1997 IEEE International Conference on Acoustics Speech and Signal Processing ICASSP 97 in Munich and a paper on this pilot study was published in the ICASSP 98 proceedings 1 Since then J DSP has been significantly upgraded with new functionality and its associated laboratory exercises are being continuously improved 2 11 J DSP is an object oriented environment that enables students to establish and run educational DSP simulations such as digital filter design signal analysis speech signal processing FFT based spectral analysis etc J DSP provides a user friend
65. ARAM gt tag in this program This is the lt PARAM gt tag in line 2 Line 2 establishes a SigGen block at the location given by the coordinates 3 1 M10 5 If we include the applet above in an HTML file and open the file with a browser the J DSP applet will be activated and by pressing Start the J DSP editor will load containing a SigGen block as shown in figure 4 Figure 4 Our first scrip loaded 3 3 Establishing connections C is used to denote connections between the blocks A lt PARAM gt tag containing C in its value will connect two blocks in the same manner as we connect two blocks by dragging from a pin of a block to a pin of another block to connect them The possible pin numbers for the blocks are shown below The format of VALUE in the lt PARAM gt tag that establishes connection between two blocks has a general format as given below C parent no parent pin no child no child pin no WV W The connection shown above requires a code given by C 0 4 1 0 This translates to connecting the first block number 0 pin 4 to the second block number 1 pin 0 If we assume this is the 6 general PARAM tag count starts from 0 the complete line is M10 6 lt param name 5 value C 0 4 1 0 gt We can now write a simple program to connect two blocks in J DSP In the code is given below observe how the lt PARAM gt tag in line 1 with numCommand as PARAM NAME has VALUE 3 because there are 3 general
66. College Committees Deans Personnel Committee member 2000 present Research Council Member 1994 1999 Engineering Excellence 2000 Committee Member November 1994 95 EE Chair Search Committee Member 1995 96 University Committees 26 Communication Advisory Committee Member 1993 96 New Courses and Course Material Developed Developed a new on line course entitled Speech Recognition ASU MEng program Developed a new on line course entitled MATLAB for DSP Applications for the ASU MEng program Developed an on line laboratory Java DSP for EEE 407 http jdsp asu edu Developed and taught a 4 Credit senior level undergraduate course in Digital Signal Processing entitled Digital Signal Processing EEE407 591 The purpose of this course is to introduce senior students to the principles and applications of Digital Signal Processing This course has become very popular among on campus and off campus students and enrollment is quite high Developed and taught a graduate level special topics course entitled Speech Coding EEE 598 The purpose of this course is to introduce to graduate students the principles and applications of speech coding Developed and taught a graduate level special topics course entitled Adaptive Filter Theory EEE 598 now established as EEE 606 The purpose of this course is to introduce to graduate students the principles and applications of adaptive filtering Developed and supervised an advanced leve
67. DI sounds Echo Generates echo effect of the input signal Reverb Generates reverb effect of the input signal Reverb Demo Simulates five specific reverb types M9 1 Block name DTMF tones Notation DTME tones Description Generates duaktone multtfrequency DTMF tones used in landline telephony applications This block generates a single tone of length 256 1 frame 1280 5 frames and 8192 32 frames samples It also generates a sequence of pre recorded tones The sampling frequency 1s 8KHz The tones can be played back using the J DSP provided sound player and used in a DSP simulation Pin assignment DTME tone Dialog window s C One C Fives ft All C Record Help Reset Java Applet Window 7 a DTMF tones dialog window Script use Name DTMF Example code lt param name 3 value BO DTME 3 1 gt Equation s Implemented y cos 2n f nT cos 27 f nT where f and f are chosen from the tone frequencies 697 770 852 941 1209 1336 1477 Hz The sampling frequency is 8 KHz 1 e T 0 125ms M9 2 M9 2 Block name MIDI Notation MIDI Description Simulates a piano keyboard and generates Musical Instrument Digital Interface MIDI sounds at the frequencies described by the MIDI standard The MIDI block can generate a single tone of length 256 1 frame 1280 5 frames and 8192 32 frames samples It can also generate a sequence of pre recorded tones The sampling frequency is 8K
68. Edit View Insert Format Tools Table Frames Window Help Type a question For help i O ad m a E Cal V d 3 A Y KO Cu H Beer i gt re lt applet CODE JDsp class width 400 height 250 gt 3 JBZ U R views _f start_jdsp html lt td width LUUS align center bgcolor Frrr lt param name numCommand value 9 gt subsp 3 lt td gt lt START PARTS gt A lt tr gt lt param name 0 value B0O siggen 1 0 gt Page lt table gt lt param name 1 value Bl plot 5 3 gt lt center gt lt param name 2 value B2 dsample 3 3 gt 4 lt div gt lt END PARTS gt lt hr SIZE 1 NOSHADE WIDTH 1005 color 000000 gt hii lt center gt lt table BORDER 0 CELLSPACING 0 Sterns Seon eee CONECTION E gt ht lt tr gt lt param name 3 value C 0 4 2 0 gt lie lt td gt lt param name 4 value C 2 4 1 0 gt Reports i lt END CONNECTIONS gt Insert Applet in body of HTML file J lt START OPEN DIALOGS gt fe lt td gt lt param name 5 value 0 1 gt RT Eoee lt END OPEN DIALOGS gt lt center gt Erp lt td gt lt part parameters gt lt tr gt lt param name 6 Hyperlinks value P0 20 10 0 0 2 a Triangular No null gt gt lt tr gt lt param name 7 value Pl b cont false gt g lt td gt lt param name 8 value P2 3 3 gt lt p align center gt lt part parameters gt Tasks
69. Hz All the tones can be used in a DSP simulation and are audible using the J DSP provided sound player Pin assignment Pin MIDI tone a e A C CL S Dialog window s MIDI tone generator l E x SOUNDS Pere rere reer reer eee eee Wee 0 Tonefone frame Tonet frames C Tonefall frames C Recording Close Update Help Reset Java Applet Window a MIDI dialog window Script use Name MIDI Example code lt param name 1 value B1 MIDI 2 1 gt Equation s Implemented y cos 2n fnT where f is taken from a MIDI standard table www mid org M9 3 M93 Block name Echo Notation Echo Description This block generates the echo effect of the input signal The echo effect is obtained by mixing the input signal with its delayed version The proportion of the delayed signal to the clean original signal determines how obvious the echo is and the delay signifies he echo period Pin assignment 1 Input time domain signal EN Output signal with echo Dialog window s Close Update Help Java Applet Window a Echo dialog window Script use Name Echo Example code lt param name 2 value B2 Echo 1 4 gt Equation s Implemented y n x n b x n R R the number of echo delay in samples In order to have a distinguishable echo R should be relatively large b is the attenuation constant Ibl lt 1 Recomended values to perceive an echo
70. Implemented N 21 h n 2 2 H k cos 2nk n a N H 0 a N 1 2 k 0 N 1 M6 9 Section M5 Frequency blocks These blocks appear at the top of the simulation area Table of blocks Block notation FFT IFFT Pk Pking Magn Phase Description Fast Fourier Transform algorithm Inverse Fast Fourier Transform algorithm Peak picking routine Calculates the magnitude of the input signal Calculates the phase of the input signal MS 1 Block name Fast Fourier Transform Notation FFT Description Implements the Fast Fourier Transform algorithm The user can select a desired FFT size Possible options are 8 16 32 64 128 or 256 Pin assignment Time domain signal x n Frequency domain signal X k 4 ee 6l S Dialog window s FFT Settings n X FFT Settings Mame E FFT Size C 8 16 C 32 64 128 0 256 Close Update Help Java Applet Window a FFT dialog window Script use Name fft Example code lt param name 3 value B3 fft 3 1 gt Equation s Implemented N 1 X k xme k 0 N 1 n 0 x n input signal X k output signal N FFT length M5 2 M5 2 Inverse Fast Fourier Notation IFFT Eee name Transform Description Implements the Inverse Fast Fourier Transform algorithm The user can select the desired inverse FFT size 8 16 32 64 128 or 256 Pin assignment Frequency domain input signal X k
71. KERSON C S BURRUS ANTHONY BESSIOS JULIE GREEN BERG N MOHANKML DALE MUGLER ERIC SOULSBY DOMINGO RODRIGUEZ PRADIP SRIMANI JOEL TRUSSELL FATIHA MERAZKA GURDAKUL PAUL HASLER C L MAX NIKIAS MASOUD SALEHI PHILIP LOIZOU SENNUR ULUKUS TAKIS KASPARIS G FAYE BOUDREAUX BARTELS BRIAN L EVANS GEORGIOS B GLANNAKIS JEFFREY FOUTZ SACHI DASH TED PAINTER GLEN P ABOUSLEMAN SASSAN AHMADI KEVIN STODDARD PANOS PAPAMICHALIS ORGANIZATION UNIVERSITY OF KENT MARQUETTE UNIVERSITY KENT UNIVERSITY STEVENS INSTITUTE OF TECHNOLOGY GEORGIA INSTITUTE OF TECHNOLOGY BLEKINGE INSTITUTE OF TECHNOLOGY BLEKINGE INSTITUTE OF TECHNOLOGY DREXEL UNIVERSITY UNIVERSITY OF NEW BRUNSWICK CANADA CAL POLY POMONA UNIVERSITY OF DETROIT MERCY UNIVERSITY OF PENNSYLVANIA NORTHEASTERN UNIVERSITY IOWA STATE UNIVERSITY RICE UNIVERSITY TEXAS INSTRUMENTS MASSACHUSETTS INSTITUTE OF TECHNOLOGY UNIVERSITY OF DETROIT MERCY UNIVERSITY OF AKRON UNIVERSITY OF CONNECTICUT UNIVERSITY OF PUERTO RICO CLEMSON UNIVERSITY NORTH CAROLINA STATE UNIVERSITY ECOLE NATIONALE POLYTECHNIQUE ALGERIA BOGAZICI UNIVERSITY TURKEY GEORGIA INSTITUTE OF TECHNOLOGY UNIVERSITY OF SOUTHERN CALIFORNIA NORTHEASTERN UNIVERSITY UNIVERSITY OF TEXAS AT DALLAS UNIVERSITY OF MARYLAND UNIVERSITY OF CENTRAL FLORIDA UNIVERSITY OF RHODE ISLAND UNIVERSITY OF TEXAS AT AUSTIN UNIVERSITY OF MINNESOTA MOTOROLA HONEYWELL INTEL GENERAL DYNAMICS NOKIA SEMY ENGINEERING TEXAS IN
72. M1 3 Note that each block has signal input s designated by the small triangular nodes on the left and signal output s to the right Some blocks carry parameter inputs and outputs at the bottom and top of the block respectively For example the Filter block has a coefficient input on the bottom and a coefficient output on the top To select a block click once to highlight it You can then move it by placing the mouse arrow over it holding down the left mouse button and dragging the box to a new location To delete a block simply select it and press the delete key on your keyboard To link blocks click once inside the small triangle on the right side of the signal generator box and while holding the mouse button down drag the mouse arrow to the triangle on the left side of the filter box Release the mouse button to create a connection between the two boxes Always make the connections in the direction of the signal flow Now connect the Coeff block that is used to specify filter coefficients to the Filter block s parameter input as shown in Fig 4 Next connect the Filter block to the Plot block so that your editor window looks like the block diagram in Fig 4 Note that you can view the dialog box of each block by double clicking on the block as shown in Fig 4 Basic blocks EL DSP P ta JE i zi G00 Pit im Aare Hanus Grape akaeesS Lats Ose Heip Filer Gaig i mY
73. M7 3 Block name Linear prediction coefficients Notation LPC Description This block calculates the linear predictor coefficients LPC The autocorrelation function is incorporated in this block in contrast to the LPC block Pin assignment Time domain signal x n LP coefficients a DE S e E Dialog window s None Script use Name LPC Example code lt param name 3 value B3 LPC 3 1 gt Equation s Implemented P Residual signal is obtained by using the equation e n x n X a n 1 i 1 l LP synthesis filter is given by H z 1 az M7 4 M7 4 Block name Lag window Notation Lag Win Description This block windows the input signal with a user defined window function The window functions available are Hamming Hanning rectangular Bartlett Blackman and Kaiser The maximum window length is 256 samples Pin assignment Autocorrelation sequence r m Windowed autocorrelation r m 6 Dialog window s Lag Window x Lag Window Name la Lag Win Type Hamming Samples 100 Window Preview 0 20 40 ED a0 100 Lag l Close i Help warning Applet Window a Lag Win dialog window Script use Name lagwindow Example code lt param name 3 value B3 lagwindow 3 1 gt Equation s Implemented r m w m r_ m r m is the autocorrelation sequence and r m the windowed autoc
74. PC Description This block computes the LP coefficients from the line spectral pairs Pin assignment 5 Line spectral pairs F 2 LP coefficients a Je E D Slo Sl Dialog window s LPC Coefficients Name Te Order 10 LPC coefficients alo 1 0 alt 1 0968 ajj 04374 aja 0 1904 ajj 05455 af5p 0 3978 ajj 0 0617 aff 0 1363 aje 0 4543 a9 0 3365 ajio 0 0194 Close Update Help Java Applet Window a LSP gt LPC dialog window Script use Name Isp2Ipc Example code lt param name 3 value B3 lsp2Ipc 3 1 gt Equation s Implemented A z where F z sum polynomial F z difference polynomial and A z LP filter F z F Z 2 M8 7 M8 10 Block name Bandwidth expansion Notation BW Exp Description This block performs the bandwidth expansion operation Pin assignment Filter coefficients Bandwidth expanded filter coefficients E e ON d Dialog window s Bandwidth X Bandwidth Expansion Exp Coef lo 9 Close Update Help Java Applet Window a BW Exp dialog window Script use Name BWExp Example code lt param name 3 value B3 BWExp 3 1 gt Equation s Implemented b b z b z7 b z Input filter transfer function H z lra Fa atay 10 _ 10 by byz by z boy z Bandwidth expanded filter H z P g l ayz ay z a y 2 where y is the ban
75. SEE Frontiers in Education conference 6 see copy of the assessment report and relevant paper in Appendix C Assessment results from this award have recently been submitted to the Online Evaluation Resource Library OERL A comprehensive manual for J DSP was also developed and posted on the internet and a hard copy of this manual is in Appendix F The free dissemination of J DSP to other institutions started in the Fall of 2002 A dissemination package was developed Appendix C and the J DSP software and labs have been disseminated to many prominent faculty members in ranked engineering programs Universities that obtained free the J DSP Version 1 executable software ISBN 0 9724984 0 0 included MIT U Penn Georgia Tech University of Minnesota Rice University and several others see complete list Appendix C 3 In the summer of 2000 NSF awarded a CCLI EMD grant that enabled us to upgrade the DSP functionality and extend the use of J DSP in other related courses such as communications image processing and controls 11 The work on the NSF J DSP project is ongoing and several new functions are being developed Such functions involve MS Internet Explorer is registered by Microsoft Incorporated MATLAB is registered by The Mathworks J DSP is sponsored by ASU and NSF CCLI EMD DUE 0089075 e J DSP Scripts enabling instructors to effortlessly create J DSP simulations and seamlessly integrate them Appendix A Fig A 3 in their web content
76. STRUMENTS ALEXANDER D POULARIKAS YAN WU JAN E ODEGARD JOHN PIERRE DYLAN DIZON OSUACDO CLUA CARLOS GODFRID M S FADALI MARK BURGE ALEXEY MATVEYEV CHRISTINA M PETERSON UNIVERSITY OF ALABAMA IN HUNTSVILLE Section M10 J DSP Scripts 1 J DSP Script basics J DSP is fitted with an interpreter designed to decode parameters contained in a simple Hypertext Markup Language HTML file which when loaded through a browser invokes the J DSP editor The editor in turn interprets the parameters and loads a J DSP flowgram as described by these parameters The parameters contained in the HTML file are written in JavaScript and are actually components of a Java applet referred here as a J DSP script This J DSP Editor capability has been designed in order to allow instructors save their own J DSP simulation examples on the Internet easily integrating interactive content in classroom web sites In addition to this introduction this manual consists of two more main sections that further discuss J DSP scripts Section 2 of this manual instructs the user on how to create J DSP scripts automatically while section 3 is an elaborate description of how to manually prepare the script code Please note that all scripts must be saved in the same directory the J DSP editor s class files are located in 2 Generating scripts automatically While trivial to prepare manually the J DSP Editor has been provided with the ability to automaticall
77. Time domain output signal x n Dialog window s IFFT Settings FFT Settings Mame lh IFFT Size Ca C16 C 32 C 64 C 128 ff 256 Close Update Help Warning Applet Window a IFFT dialog window Script use Name ifft Example code lt param name 3 value B3 1fft 3 1 gt Equation s Implemented N I x n P x4 yer Nn 0 N 1 k 0 X k input signal x n output signal M5 3 MS5 3 Block name Peak Picking Notation PkPking Description Selects a specific number of peaks from a frequency domain signal The first set of peaks or the highest magnitude ones can be selected Here the Peaks selected option allows users to specify how many peaks to be selected For example 64 is chosen in the graph below In this case the First option selects the first 64 peaks of the input signal and the Highest option selects the 64 peaks that are the larger in magnitude Pin assignment ra Dialog window s Peak Picking x Peak Picking Hame ji 30 MM agnitude Linear Hlug FFT gray peaks selected Signal size 256 Peaks selected E4 C Fist amp Highest Close Update Help Warning Applet Window a PkPking dialog window Linea dB Script use Name peakpicking Example code lt param name 3 value B3 peakpicking 3 1 gt M5 4 MS 4 Block name Magnitude Notation Magn Description This block calculates
78. W B Mikhael and A S Spanias Accurate Representation of Time Varying Signals using Mixed Transforms with Applications to Speech IEEE Trans on Circuits and Systems Vol 36 No 2 pp 329 331 Feb 1989 W B Mikhael and A S Spanias Efficient Modeling of Dominant Transform Components Representing Time Varying Signals IEEE Trans on Circuits and Systems Vol 36 No 2 pp 331 334 Feb 1989 W B Mikhael and A S Spanias A Fast Frequency Domain Adaptive Algorithm Proc of the IEEE Vol 76 No 1 pp 80 82 Jan 1988 W B Mikhael and A S Spanias Comparison of Several Frequency Domain LMS Algorithms IEEE Trans on Circuits and Systems Vol 34 No 5 pp 586 588 May 1987 W B Mikhael A S Spanias G Kang and L Fransen A Two Stage Pole Zero Predictor IEEE Trans on Circuits and Systems Vol 33 No 3 pp 352 354 March 1986 S Miller and A S Spanias Antenna Beamforming using Adaptive Quiescent Pattern Control Submitted to EURASIP Journal on Applied Signal Processing EURASIP JASP Special Issue on Smart Antennas Submitted May 2003 S Miller and A S Spanias N Chakravarti A Spanias L D Iasemidis and K Tsakalis AR Modeling of DNA sequences Submitted to EURASIP Journal on Applied Signal Processing EURASIP JASP Special Issue on Genomic Signal Processing Submitted Feb 2003 Refereed Papers in National and International Conference Proceedings C 1 C2 C 3 C 4 C5 C 6 B Ba
79. ab 4 FIR and IIR filters Q1 The contents of this exercise helped you understand the concepts of FIR and IIR filter design L Strongly Agree Agree Neutral Disagree nann Strongly Disagree Q2 Which part of the exercise helped you the most to understand the concepts of FIR and IIR filter design E Q3 Performing the exercise can you decide which window to use for sharp transitions in a filter L Strongly Agree Agree Neutral Disagree nnan nA Strongly Disagree Q4 Through the IIR filter exercise you know which IIR filters give a monotonic pass band L Strongly Agree Agree Neutral Disagree nnana Strongly Disagree Q5 Did you experience any problems in terms of connection time to implement etc Please describe Q6 Was there enough information in the help screens to assist you in using the blocks La Yes L No La Did not use the help screens Q7 Is there any particular concept which you could not grasp earlier from your text book class but became clear from this exercise If so please describe El Q8 Performing the exercises you are now more comfortable with these topics s Yes s No Q9 Setting up the required lab simulations was pretty easy L Strongly Agree Agree Neutral Disagree nnan n Strongly Disagree Q10 Can you suggest an exercise along the lines of this one Q11 Please suggest possible improvements relative to this lab such as redesigning of dia
80. and T Painter An Educational Software Tool for the Study of Speech Coding Algorithms in a DSP Class Special Issue on DSP Education IEEE Trans on Education Vol 39 pp 143 152 May 1996 K Tsakalis M Deisher and A Spanias System Identification Based on Bounded Error Constraints JEEE Transactions on Signal Processing Vol 43 No 12 pp 3071 3075 Dec 1995 J Liu A Spanias and J Maisel A Real time Adaptive Interference Canceller using the BLMS Algorithm American Society of Engineering Education ASEE J Eng Tech Vol 12 No 1 pp 34 38 Spring 1995 P Loizou M Dorman and A Spanias Automatic recognition of syllable final nasals preceded by e Journal of Acoustical Society of America Vol 97 3 pp 1925 1928 March 1995 A S Spanias Speech Coding A Tutorial Review Proceedings of the IEEE Vol 82 No 10 pp 1441 1582 October 1994 M Deisher and A S Spanias Practical Considerations in the Implementation Frequency Domain Adaptive Noise Cancellation IEEE Transactions on Circuits and Systems Part Il Analog and Digital Signal Processing Vol 41 No 2 pp 164 168 Feb 1994 A Spanias M Deisher P Loizou G Lim and B Mears A New Highly Integrated Architecture for Speech Processing and Communication Applications Intel Technology Journal Special Issue on Computer Supported Cooperation pp 41 56 Spring 1994 A S Spanias IEEE Transactions on Signal Processing
81. and digest several topics that were not immediately obvious by attending the class or reading the text book 7 1 3 J DSP Learning objectives and ABET accreditation criteria We have carried a self study on EEE 407 and re evaluated how the course and particularly its J DSP laboratory addresses ABET criteria With regard to ABET design requirements students are asked to solve open ended problems as part of the J DSP laboratory exercises The J DSP lab involves extensive filter design using several methods and students are asked to analyze these filter design methods and perform comparisons Students are asked to design algorithmic steps and develop J DSP realizations of these algorithmic steps More specifically the J DSP laboratory component addresses well ABET Criteria 3 and 8 The J DSP laboratory tasks in almost all the J DSP labs described in Appendix D engage the students into applying their knowledge of mathematics and engineering ABET 3a For example in J DSP Labs 2 and 5 students apply the mathematical properties of Fourier transforms and z transforms to real world engineering applications such as filtering and spectral analysis relates to ABET 3a There are specific labs that involve filter design J DSP Lab 4 in Appendix D where the student is required to use J DSP to perform several designs and compare and choose the one that satisfies specific criteria ABET 3b ABET 3c relates to J DSP lab design tasks such as the design of linear phase f
82. andard Deviation o Total energy x n Power x x n n l n l x n input signal N number of samples M2 10 M2 8 Block name Plot 2 Notation Plot2 Description Plots two signals in the same dialog window All signals are plotted in terms of samples and any scale changes apply to both graphs Graphs can be plotted one below the other one next to the other or in the same axis Use the Graph Position option to vary the graph location Pin assignment Input signal x n Input signal y n Output signal z n Input signal x n Output signal g n Input signal y n a Dialog window s x Mame fe Graph Position ae Magni scale linear C dB 1 0E0 1 0E Pog 13 0 6 32E 1 0 0 0 0 1 0E0 1 0E0 o tire samples 30 oO tire samples 30 Grid SameXaxis Plot cont Axis Auto Close Help Java Applet window a Plot 2 dialog window Horizontal orientation M2 11 Mame fk Graph Position Magn scale linear dB 1 020 0 0 1 0E0 tire samples 30 1 0E0 0 0 1 0E0 tire samples 30 Grid Same axis Plot cont Axis JAuto Close Help Java Applet windo b Plot 2 dialog window Vertical orientation Plot Name d Graph Fosition iE Amplitude scale linear dB Close Help Java Applet Window c Plot 2 dialog window same axis option Script use Name plot2 Example code lt param name
83. architecture for speech processing and communication applications Intel Technical Journal Special Issue on Computer Supported Cooperation pp 41 56 Spring 1994 4 S Bellofiore J Foutz C Balanis A S Spanias J Capone and T Duman Smart Antenna System Analysis Integration and Performance for Mobile Ad Hoc Networks MANETs Full paper to appear in IEEE Trans Antennas Propagation Special issue on Wireless Communications vol 50 no 3 March 2002 results of 2 year NSF ITR project A Spanias Co PI 5 S Ahmadi and A Spanias Algorithms for Low bit rate sinusoidal coding Speech Communications Vol 34 2001 pp 369 390 June 2001 A Spanias principal investigator and Ph D advisor 6 Min Tau Lin A S Spanias and P Loizou Speech Recognition Using Minimum Error Classification Speech Communication vol 30 pp 27 36 January 2000 A Spanias principal investigator and Ph D advisor 7 S Ahmadi and A S Spanias Cepstrum Based Pitch Detection Using a New Statisctical V UV Classification Algorithm IEEE Trans on Speech and Audio Vol 7 No 3 pp 333 338 May 1999 A Spanias principal investigator and Ph D advisor 8 P Loizou and A S Spanias Improved speech recognition using a subspace projection approach IEEE Trans on Speech and Audio Processing vol 7 no 3 pp 343 345 May 1999 A Spanias principal investigator and Ph D advisor 9 S Ahmadi and A S Spanias
84. as used several times in class sessions I plan to turn this in to a book Developed teaching material consisting of 250 viewgraphs for the speech coding course EEE 598 This was made available to students along with a 100 page report developed by Andreas Spanias It was used consistently in class sessions This material along with software is being developed as a text book expect completion next year a Undergraduate Projects Supervised Java Software for introducing undergraduatres to DSP Carolyn Cooper Spring 2003 This project satisfies the requirements of the EEE 490 course Array Microphones Steve Brown Spring 2001 This project satisfies the requirements of the EEE 490 course Real time Spectral Analyzer on the DSP 56000 Students Francine Doyle and Umberto Santoni The project entailed programming FFTs in assembly language and developing a filter bank scheme using the theory of the Discrete Fourier Transform The software provides plots on the PC demonstrating time varying spectral estimates Sponsored by Motorola by donation of a DSP board This project satisfied the requirements of the EEE 490 course Analysis Compression and Synthesis of a Speech Signal Student Anastasios Policarpou and K Gharib The project entailed development of assembly code for the short term linear prediction algorithm embedded in the GSM cellular standard The project was sponsored in part by Intel donation of the EP evaluation board
85. ate the facility to save a workspace in J DSP editor for the future use This problem is now solved as in the new version of J DSP there is a facility to import export the work space as a script file Also students reported several software bugs that have been fixed We are especially appreciative of the ability to get immediate feedback from the students on the operation of the software 17 7 CONCLUSIONS Different types of assessment instruments have been prepared and disseminated to student users in the ASU EEE 407 class in the Fall 2002 and Spring 2003 Continuous feedback helped correct and improve the J DSP software and exercises The students in general found the J DSP concept very convenient and easy to use Concept specific assessments and pre post assessment revealed that several J DSP functions have been particularly useful in communicating key DSP concepts J DSP was proven to be particularly useful in learning issues related with filter design and interpretation of frequency spectra The J DSP visualizations involving pole zero diagrams have shown prominent differences in pre and post assessments that lead us to believe that we need to integrate even more animation and develop demonstrations that are dynamic In all EEE 407 students asserted that they have benefited from J DSP and they particularly appreciated the fact that the tool was available on the web from any location Industry students taking the course from remote sites have
86. ation e to expose students to filter banks of the type used in MP3 and other compression formats e to expose students to random signal analysis by simulating filters with random inputs e to expose students to periodograms correlograms and linear prediction e to provide students with an introduction to speech processing 4 General J DSP Labware Description and Usage The software is used to support the laboratory portion of EEE 407 Laboratory exercises sample copy in Appendix D are assigned on a weekly basis These laboratory exercises actually reside on the WWW While access to the J DSP program is free and universal access to the laboratory page is secure because it invokes student information grades etc Students perform EEE 407 labs as follows The student follows instructions on the secure EEE407 laboratory web page and starts the J DSP program by pressing a button that activates the software Before starting J DSP students are reminded to download the Java virtual machine link provided on the web that enables the MS Internet Explorer browser to run J DSP The student then proceeds to form and execute the required J DSP simulations generate graphs and data and then fill out an electronic report form The electronic report allows the students to answer questions attach J DSP data and graphs provide comments and textual interpretation of results and respond to an electronically graded quiz This form is processed by servlets d
87. atistical DSP blocks These blocks appear at the top of the simulation area Table of blocks Block notation Description Autocorr Computes the autocorrelation values of a signal LPC Computes the linear predictor coefficients LPC LPC Computes the linear predictor coefficients LPC Lag Win Windows a time domain signal Sym Corr Finds the symmetric autocorrelation Corlogrm PSD estimation using Correlogram method Prdogrm PSD estimation using Periodogram method Spectrogram Provides frequency versus time plots AR Est AR estimation based on the Levinson Durbin algorithm Autocorr ec f Lec Lagwin symCorr corogrm Prdagrm J Spectrogram AR Est_ M7 1 Block name Autocorrelation Notation Autocorr Description This block calculates the autocorrelation sequence of a signal The user needs to specify the number of lags and select whether they are computed for a particular frame this frame or for all frames An option for biased or unbiased normalization is provided Pin assignment Time domain signal x n Autocorrelation 7 m Dialog window s a 0 00433 a 0 0036247 0 001767 3 36E 4 0 00207 26 0 0031334 0 0032871 0 0024403 8 669E 4 8 347E 4 10 0 0020512 11 0 0024256 Autocorrelation Mame fa i Frame Size 256 Samples No of Autocar 100 Lags 0 1 2 3 4 5 6 7 8 9 Compute Lage for This frame All fames F Normalization 0 20 40 6
88. attenuating the noise components between the harmonics and by exploiting the noise masking effects of the sinusoids He also developed s new segmentation technique 12 and other auxiliary algorithms that improve alphabet recognition 6 8 11 Andreas Spanias published comprehensive survey papers on audio coding 1 and speech coding 2 in the Proceedings of the IEEE As evidence of the quality of his work it is noted that publication 1 received the prestigious 2002 Donald Fink IEEE wide paper prize award Andreas Spanias collaborates extensively with industry He has established and directed a five year 1 5 million dollar ASU program funded by Intel Corp towards the design of the low power DSP core 60172 and its multiprocessor version named Phoenix architecture His work involved chip design and analysis He made detailed design recommendations 3 on the arithmetic unit instruction set architecture interface and on chip memory to accommodate low power implementation of wireless telecom standards involving source and channel coding as well as other physical layer functions He also provided a detailed analysis on the numerical behavior of the chip and developed algorithms to handle numerical problems in the fixed point realization of the source and channel coders The two papers of Andreas Spanias in the Proc of the JEEE 1 2 provide a comprehensive survey of state of the art algorithms and research in speech and audio coding Paper 1 received a
89. cepts of the Z transform Improvement of your understanding of the concepts of pole zero plots and frequency response Improvement of your understanding of the concepts of FIR and IIR filter design Improvement of your understanding of the general concepts of using FFT in signal analysis You have learned how to generate a sinusoid with a digital filter You have learned which window to use for sharp transition in a filter from lab 4 The exercise helped you to clearly visualize signal symmetries on the FFT spectra EN 50 47 42 24 29 42 18 46 44 47 6l 55 47 65 13 11 16 TABLE Ul STATISTICS BASED ON USER EVALUATION YES NO OF LAB 1 2 3 AND 4 Lab Yes Evaluation questions No No 1 Understand more clearly the relationship of the impulse response with the transfer 1 95 2 function 2 Understand more clearly that spectral resolution of the FFT is limited by frame size 4 99 1 window type and window size i 56 17 2 33 17 3 Enough information in the help screen 3 40 23 4 47 13 1 93 7 4 Performing the exercises you are now more 5 Ji comfortable with the topics related with each m lab assignment 4 90 10 rest 27 30 37 and 40 did not use help screen in Lab 1 2 3 and 4 respectively Students also recommended some of the changes in J DSP on line tools used in the lab simulations One common suggestion is to accommod
90. cessor A Spanias and M Deisher To Intel MRC ASU TRC Technical Report TRC SP ASP 9501 May 1995 R 5 Preprocessing Algorithms for Speech Recognition A Spanias P Loizou and G Tucker ASU TRC Technical Report TRC SP ASP 9403 May 1994 R 6 Speech Coding Algorithms for Mobile Communications A Review A Spanias ASU TRC Technical Report TRC SP ASP 9402 April 1994 R 7 Speech Enhancement for Mobile Communications A Spanias and M Deisher ASU TRC Technical Report TRC SP ASP 9401 Reported to Intel Corporation April 1994 R 8 Implementation of the VSELP Algorithm on the EP Addendum to Fixed point Implementation of the VSELP Algorithm A Spanias and M Deisher ASU TRC Technical Report TRC SP ASP 9305 Reported to Intel Corporation October 1993 R 9 Speech Enhancement Using the Pseudocepstrum Final Report A Spanias and K Daroudi ASU TRC Technical Report TRC SP ASP 9304 Reported to Motorola Inc June 1993 R 10 Design Analysis and Implementation of the GSM Modem A Spanias Y Zhang and F Tiong ASU TRC Technical Report TRC SP ASP 9303 Reported to Intel Corp March 1993 R 11 Active Noise Cancellation in Ducts Final Report A Spanias and J Liu ASU TRC Technical Report TRC SP ASP 9302 Reported to ANVT March 1993 R 12 Speech Enhancement Using the Pseudocepstrum Task 3 A Spanias and K Daroudi ASU TRC Technical Report TRC SP ASP 9301 Reported to Motorola Inc March 1993 R 13 Speec
91. cognition Technologies A Review Proc IEEE International Symposium on Circuits and Systems ISCAS 91 Invited pp 572 577 Singapore June 1991 A S Spanias and F H Wu Speech Coding and Recognition A Review Proc of the First Cyprus International Conference on Computer Applications to Engineering Systems Invited pp 46 71 July 1991 S B Jonsson and A S Spanias Seismic Data Compression IEEE International Phoenix Conference on Computers and Communications IPCCC 90 Conf Proc Invited pp 276 279 Phoenix March 1990 A S Spanias and W B Mikhael An Adaptive Bit Allocation Scheme for Coding of Speech Signals Using Partial Sets of Orthogonal Functions 32nd Midwest Symposium on Circuits and Systems MWCAS 89 Conf Proc MWCAS 89 Invited Session TAM4 S Champaign Illinois August 1989 14 C 86 W B Mikhael and A S Spanias Representation of Speech Signals using Mixed Incomplete Sets of Basis Functions 21st Asilomar Conference on Circuits Systems and Computers Conf Rec Vol 2 Invited pp 905 908 Pacific Grove California Nov 1987 C 87 W B Mikhael A S Spanias and F H Wu ARMA modeling by cascading a Linear Predictor and a Pole Zero Structure ISTh Asilomar Conference on Circuits Systems and Computers Conf Rec Invited pp 63 70 Pacific Grove California Nov 1984 C 88 A Clausen T Painter A Xavier M Tampi T Lam A Constantinou and A Spania
92. ction We observe that 92 students answer correctly before performing the lab and 92 3 students answer correctly after they have finished the lab Percentage improvement is negligible as the question was evidently very simple In Question 2 students were asked to find the poles and zeros of a given transfer function 88 students answered correctly before they started working on the lab and 92 3 students answered correctly after completing the lab In Question 3 we asked students to find the impulse response for a given transfer function that is sum of two first order all pole filters i e the composite system consists of two parallel systems 76 students answered correctly before they attempted the J DSP Lab 2 assignment and 94 9 students answered correctly in the post lab assignment The percentage improvement is 18 9 and is noticeable g Pre assessment Assessment of LAB 1 w Post assessment O Improvement 0 N DO n O O ao c hn A students who answer correctly Q1 Q2 Q3 Q4 Q5 Q6 Question Number FIGURE 4 A THE RESULTS OF LAB 1 ASSESSMENT In Question 4 a sinusoidal impulse response of a system 1 e a digital oscillator is given and the students are asked to choose the true characteristics of the filter out of four choices 17 5 more students answered correctly after using J DSP In Question 5 students are asked 1f it is possible to suppress completely a sinusoid of a certain f
93. d they are merely code to establish the simulation These scripts activate the J DSP applet containing the Start button By pressing this button the programmed flowgram will appear To create J DSP scripts an HTML parameter tag known as the lt PARAM gt tag is used The use of this tag is described below The general format for a lt PARAM gt tag is lt PARAM NAME parameter1Name VALUE a Value gt The first line of the J DSP script is a lt PARAM gt tag which specifies the total number of J DSP script lines The parameter name is numCommand and its value is the number of lt PARAM gt M10 4 tags that will be used in this particular applet excluding itself This will become clearer as the lt PARAM tag description continues The format for the first line of the script is lt PARAM NAME numCommand VALUE number of param tags below gt For the general lt PARAM gt tags that follow each tag s PARAM NAME is given a unique number This number increases sequentially starting from 0 For example lt PARAM NAME 0 VALUE a Value gt lt PARAM NAME 1 VALUE another Value gt The PARAM NAME VALUE etc keywords are case insensitive so it is not necessary to use capital letters In the VALUE part of the lt PARAM gt tag the parameters B C O P and are used to specify different instructions For example B is used to instruct the editor to create a new part block These parameters are further e
94. dke and A Spanias Partial Band Interference Excision For GPS Using Frequency Domain Exponents IEEE International Conference on Acoustic Speech and Signal Processing ICASSP 2002 Vol 4 pp 3936 3939 Orlando May 2002 N Chakravarti A Spanias L D Iasemidis and K Tsakalis AR Modeling of DNA sequences 22 IASTED International Conference MIC 2003 538 542 Innsbruck Austria February 10 13 2003 T Thrasyvoulou K Tsakalis and A Spanias J DSP Control A Control Systems Simulation Environment 2 TASTED International Conference MIC 2003 538 542 Innsbruck Austria February 10 13 2003 Balaji Veeramani K Narayanan Awadhesh Prasad A Spanias and L D Iasemidis On the use of the directed transfer function for nonlinear systems Proceedings of IASTED International Association of Science and Technology for Development International Conference Palm Springs California USA Feb 24 26 2003 pp 270 274 Balaji Veeramani Awadhesh Prasad K Narayanan A Spanias and L D Iasemidis Measuring information flow in nonlinear systems A modeling approach in the state space Proceedings of the 40th Annual Rocky Mountain Bioengineering Symposium Biloxi Mississipi USA April 11 13 2003 in press Rajeshkumar Venugopal K Narayanan A Prasad A Spanias J C Sackellares and L D Iasemidis A new approach towards predictability of epileptic seizures KLT dimension Proceedings of the 40th Annual Rocky
95. dwidth expansion coefficient M8 8 M8 11 Block name Inverse Transfer Function Notation Inv TF Description This block inverts the transfer function at its input Pin assignment Filter coefficients Inverse transformed transfer function E a E Dialog window s None Script use Name Inv TF Example code lt param name 3 value B3 Inv TF 3 1 gt Equation s Implemented b b z b z bz Input filter H z U lra Fa e Fa a z a z az b Inverse transformed transfer function H z F 5 ET LFO FOZ nO eZ Oe M8 9 M8 12 Block name Perceptual weighted filtering Notation Prep Fil Description This block performs the perceptual weighted filtering or simply perceptual weighting The weights 1 2 can be entered by the user Pin assignment LP coefficients A z Perceptual weighted output W z lt A ee ES Dialog window s Perceptual x Perceptual Weigthing Filter Gamma 1 0 9 Gamma 2 0 6 Close Update Help Java Applet Window a Prcp Fil dialog window Script use Name Prep Fil Example code lt param name 3 value B3 Prcp Fil 3 1 gt Equation s Implemented 10 l l A z Y 1 y az Perceptual weighting filter is given by W z AAH i j l A z Y gt Le i i I Y 4a z 3 gt Yy Y are the perceptual weights and a are the LP coefficients M8 10 Section M7 St
96. e FIR design Notation FIR Description Designs a finite impulse response FIR filter based on the windowing method The windowing FIR filter design method is a straightforward technique implemented by expanding the frequency response of an ideal filter in a Fourier series and then truncating and smoothing the response using a window The user needs to supply the following information Window type Hamming Hanning Blackman Bartlett rectangular or Kaiser Filter order maximum is 64 Type low pass high pass pass band or stop band Cut off frequences f take values from O to 1 where fe corresponds to half the sampling frequency Pin assignment Filter coefficients Es ee Dialog window s FIR Filter Design based on the window method FIR Filter Parameters Name ja Window Type Hamming Order e Type LowPass z Cut off Freg po a2 COEFFICIENTS 0 00506 fe Close Update Help Passhand Transition band Stopbhand Warning Applet Window a FIR dialog window and filter design specifications Script use Name FIR Example code lt param name 3 value B3 FIR 3 1 gt M6 4 M6 4 Block name IIR design Notation IIR Description Designs an infinite length impulse response IIR filter based on the bilinear transformation Butterworth Chebyshev I amp II and Elliptic filters are supported The filter specifications are in terms of
97. e Help Warning Applet window a Correlogram dialog window Script use Name corrlog Example code lt param name 3 value B3 corrlog 3 1 gt Equation s Implemented N1 _ j2tkm R k H Ee mje J N the length of the sequence M7 7 M7 7 Block name Periodogram Notation Prdogm Description This block estimates the power spectral density PSD by operating directly the data set Two different periodograms can be used to estimate the PSD sample spectrum or Welch periodogram The user can specify the number of smooth over points to implement the Daniell periodogram over the sample or the Welch periodograms Pin assignment i Dialog window s Pariadcayrained 440 11g A ATA OLBo R280 n dg Ti Wagriude rda Fd 09410 464 Mame f t L 5 AFT 1929169 160731201045 ATG AT Haar fare AFL O36 149084 Frame Sipe 1256 7 TE ORE Baie 158 abe FFT Ske 256 ATIJ HAm Peiodipam Sample Spectrum aei E h a a PANGIN jRectangubsr Fan 1A gidir fe 2 170 0565644 54 moih oar F Fainte 3 10 527736600 gt Cee Values Liedale Help Cine a Prdogm dialog window and output values Script use Name periodgm Example code lt param name 3 value B3 periodgm 3 1 gt Equation s Implemented _ j2mkn 2 1 N 1 The sample spectrum of the p frame is given by R k a X w n x ne NY n 0 P Welch periodogram R
98. e Sinusoidal Speech Coding pp 345 349 IEEE International Symposium on Circuits and Systems 2000 ISCAS 00 Session P2 P1 Geneva Switzerland May 28 31 2000 A Spanias S Urban A Constantinou M Tampi A Clausen X Zhang J Foutz and G Stylianou Development and Evaluation of a Web Based Signal and Speech Processing Laboratory for Distance Learning Proc IEEE International Conference on Acoustic Speech and Signal Processing ICASSP 2000 Istanbul June 2000 Andreas Spanias Argyris Constantinou Jeff Foutz and Fikre Bizuneh An on line signal processing laboratory IEEE DSP Education Workshop Hunt Texas October 15 18 2000 K Daroudi and A Spanias Frequency Selective Adaptive Modeling pp 333 337 Proc MIC 2000 Innsbruck Feb 14 17 2000 Hiren Bhagatwala Edward Painter and Andreas Spanias An Interactive GUI based Tool for Signal and Speech Processing Courses Proceedings of ICASSP 99 Phoenix March 1999 Axel Clausen and Andreas Spanias An Internet based Computer Laboratory for DSP Courses Proceedings of the ASEE IEEE Frontiers in Education Conference Tempe November 1998 Xavier Anand Andreas Spanias and Ted Painter An Adaptive System Identification Java Simulation for Internet based Software Proceedings of the ASEE IEEE Frontiers in Education Conference Tempe November 1998 A Clausen A Spanias A Xavier M Tampi A Java Signal Analysis Tool
99. e domain signal x n 2 LPC spectrum R x k Ea AR coefficients a Dialog window s AR Spectral Estimation Name Lag Window Rectan gular AR Order aorta 020083 4 255 4 1275 4 2 1058 0 16577 41 0852 Oita 4 14628 itik Normalization Unbiased Biased eff Close Update Help a AR Est dialog window Java Applet Window Script use Name AREst Example code lt param name 3 value B3 AREst 3 1 gt Equation s Implemented Rie k 1 X az i Here a Linear Prediction LP coefficients and N is the order of the LP filter M7 10 Section M6 Filter blocks These blocks appear at the top of the simulation area Table of blocks Block notation Description PZ Placement Allows entering pole zero values PZ Plot Plots poles zeros in polar coordinates FIR Design FIR filter design IIR Design IIR filter design Kaiser Kaiser filter design Parks McClellan Parks McClellan filter design LMS LMS adaptive filter algorithm Freq Samp Frequency sampling PZPlacement PZPlot FIR Design 1R Design Kaiser Design Parks MeClellan Lms Freq Sampling M6 1 Block name Pole Zero Placement Notation PZ Placement Description This block allows the user to enter poles and zeros representing a filter The corresponding filter coefficients are passed to the output Poles and zeros are added as conjugate pairs and no more than 10 5 pai
100. e simulations and establishes a new paradigm for running DSP simulations and laboratories over the Internet e Prof Antonia Papandreou Suppappola of Arizona State University writes The J DSP software provides a unique active learning methodology for digital signal processing As a result we started to use some of its functionalities in other Systems area courses such as Communications Systems and Advanced Signal Processing courses e Professor Tolga Duman of Arizona State University writes The students and the instructors in this course believe that this on line tool is extremely important for student learning specifically student evaluations of the software have been consistently positive e Professor G Faye Boudreaux Bartels of the University of Rhode Island writes Z have experimented with your J DSP laboratories for senior undergraduate and graduate students and found them very useful e Professor Brian Evans from the University of Texas Austin writes your original object oriented data flow Java DSP environment will be extremely useful in enabling the students to explore and associate various important engineering applications with the abstract mathematical concepts taught in these junior level courses Complete recommendation letters and agreements to host and use J DSP are given in the Appendix B 6 2 Endorsements from Professional Societies The IEEE Phoenix section awarded Professor Spanias with the Educat
101. eb site is set up to guide them through a J DSP exercise and then an on line evaluation and assessment e High School Students We have recently developed J DSP functions to introduce high school students to DSP and Multimedia technologies 7 e Colleagues at other Universities Formal dissemination to colleagues at other universities started in the Fall of 2002 We have several colleagues that obtained copies of J DSP Appendix C 3 3 Learning Objectives 3 1 General objectives of J DSP and the associated J DSP laboratory exercises are e To provide on line laboratory experiences to undergraduate DSP students in Electrical Engineering and enable programmable Internet simulations with embedded animations e To accelerate leaning by exposing students to hands on manipulation of signals and DSP systems e To provide DSP problems and design exercises that reinforce the theory learned in class and provide intuition and complementary information usually not available in lectures and text books 3 2 Specific J DSP objectives relating to specific DSP topics are e to provide hands on experiences by simulating DSP systems filters FFTs etc e to provide visualization examples relating time domain z domain and Fourier domain e to provide hands on experiences with FIR and UR digital filter design e to familiarize students with the use of windows and FFTs in spectral analysis e to expose students to quantization effects via J DSP on line simul
102. eff Junction Filter Freq Resp Plot Plot2 Snd Player Quantizer Description Generates signals of length up to 256 samples Generates signals of more than 256 samples Allows entering numerator denominator coefficients for filters Routes its input to its two outputs Filters input signal based on provided coefficients Calculates and displays frequency response of a filter Plots a single signal Plots two signals for comparison purposes Performs signal playback Performs signal quantization See next page for button diagram Sig Gen SigGen L Junction Filter Freq Resp ond Player Cluantizer M2 2 M2 1 Block name Signal generator Notation SigGen Description Generates a variety of time domain signals It supports pulses triangular delta exponential sinusoid sinc random and user defined signals The length of each signal pulse width and the amplitude of the signal gain can be set A signal can be made periodic if the periodic option is selected The base of the exponential can also be varied Random signals can have uniform normal and Rayleigh distributions with variable mean and variance Pin assignment Pin Time domain signal ES Dialog window s Signal Generator Signal Generator Name E signal Preview Signal Sinusoid 1 Freq 9 Triangular Gain 1 Exponential sinusoid Sime Pulsewidt Random user defined 5 Peric Period i 1
103. enter ASU Speech Processing Algorithms Development Analysis and Evaluation April 21 1992 Speaker at the Motorola Group SABA Meeting at the Center for Professional Development ASU Spectral Analysis of EMG Signals July 1 1992 Seminar given at the Cyprus Institute of Neurology and Genetics CING 18 RESEARCH GRANTS AND CONTRACTS External Research Grants and Contracts L 10 11 12 13 PI Andreas Spanias CO PIs T Duman A Papandreou K Tsakalis L Karam Java DSP Extensions to Communications Advanced DSP Controls Image NSF JRA 0001 424 770 Jan 2001 Jan 2004 CO PI A S Spanias PI C Balanis and 4 other CO PIs NSF Smart Antennas 458 100 Sept 2000 Aug 2002 PI A S Spanias Intel Corp Distributed Voice Recognition System for the PC 58 100 DWT0018 Sept 1996 Jan 1998 PI A S Spanias and CO PI J Sadowsky Analysis and Implementation of CDMA Mobile Communications Amount 241 457 00 Intel Corp DWT 0011 Aug 1996 Aug 1997 PI A S Spanias Development of Universal and Interoperable Speech and Audio Compression Algorithms for Multimedia and Teleconferencing Applications Sponsor Intel Corp Amount 177 354 DWT 4598 Feb 1995 Jan 1998 PI A S Spanias and CO PI J Sadowsky Implementation and Integration of the Speech Codec Channel Coder Decoder and Signaling Protocol on Prototype DSP Chips Intel Corp Amount 243 500 00 DWT 4630 May 199
104. es Organizaer and Chair Mini workshop on Signal Processing for Communications and Multimedia Tempe Feb 2002 Program Committee Melecon 2000 10th Mediterranean Electrotechnical Conference May 29 31 2000 CYPRUS Organizer and Chair Special Session on Speech Coding 1995 International Conference on DSP Limassol 1995 Program Committee and Session Chair 1995 IEEE Midwest International Symposium on Circuits and Systems MWCAS 95 Brazil August 1995 Co Organizer and Chair Special Session on Recent Advances in Speech Processing 1991 IEEE International Symposium on Circuits and Systems ISCAS 91 Singapore June 1991 Organizer and Chair Special Session on Low rate signal coding 1990 IEEE International Phoenix Conference on Computers and Communications IEEE IPCCC 90 Scottsdale March 1990 Track Chair for the Communications Technology Sessions 6 sessions of IEEE IPCCC 91 Session Chair in the IEEE PCCC 92 and IPCCC 93 Session Chair in the 1993 International Conference on DSP and CAES Nicosia 1993 University Liaison and Program Committee member in IEEE IPCCC 92 IPCCC 93 and IPCCC 94 Co organizer Special Session on Representation of Time Varying Signals 21st Annual Asilomar Conference Asilomar Conference Grounds Pacific Grove Nov 1987 25 Professional and Scientific Service cont Paper and Book Reviews Reviewed papers for several IEEE Transactions 1 e Circuits and Systems Signal Processing Communicatio
105. escribed in reference 2 which produce an electronic HTML report for each student that compiles the student responses graphics etc The instructor then evaluates and grades the report electronically An electronic grade book is interfaced with the J DSP labware Specifics on the report submission are given in reference 2 and samples of reports are given in Appendix D 3 4 1 Description of a Sample J DSP Laboratory Exercise and the Associated Learning In J DSP Lab 2 The z transform and Frequency Responses Appendix D 1 e J DSP Problem 2 1 The student is required to design and simulate in J DSP a discrete time system with an exponential impulse response LEARNING The student learns how to use J DSP to invert z transforms and obtain computationally the time domain signal The J DSP visualization exercise reinforces the relation of z plane singularities with filter stability e J DSP Problem 2 2 The student is required to design and simulate using J DSP a digital oscillator LEARNING The student learns how digital oscillators are designed and implemented Digital software oscillators are used in cell phones to generate the dialing DTMF tones e J DSP Problem 2 3 The student is required to design and simulate using J DSP an FIR filter that cancels sinusoidal interference LEARNING The student learns how an FIR filter can be configured to cancel 60Hz interference e J DSP Problem 2 4 The student is required to simulate a linear phase
106. essed signal SS S Ee ee Ce Ed EA Dialog window s Signal to Noise Ratio SNR Eg Name f SNR 10log Fs7 n s n s n F s n reference signal s n processed signal Help Close Java Applet Window a SNR dialog window Script use Name snr Example code lt param name 3 value B3 snr 3 1 gt M3 3 M3 3 Block name Statistics Notation Statistics Description This block computes the first order statistics of the input signal i e the mean the variance and the standard deviation The mean is calculated as the sum of the individual samples of the input divided by the number of samples The variance is a measure of the ceviation from the mean Standard deviation is the square root of the variance Pin assignment Dialog window s Statistics Mame E Wean 0 0618 Varlance 1 32183 Std Dew 1 14971 Help Close Java Applet Window a Statistics dialog window Script use Name stats Example code lt param name 3 value B3 stats 3 1 gt Equation s Implemented 1 _ Mean u roA x n Variance o N X x n uU Standard deviation n 1 n 1 x n input signal of length N M3 4 M3 4 Block name Window Notation Window Description This block performs a windowing operation on the input signal The available window functions are Hamming Hanning rectangular Bartlett Blackman and Kaiser The maximu
107. ethod The design process involves calculating the Fourier series of the ideal filter and then multiplying it with a Kaiser window that best fits the filter specifications Filter specifications are Filter type can be low pass high pass stop band or pass band Wp Ws pass band and stop band edge cut off frequencies respectively Wp2 Ws2 second pass band and stop band edge cut off frequencies respectively for pass band filters PB SB pass band and stop band tolerances in dB Pin assignment Filter coefficients Es ee Dialog window s Kaiser FIR Filter Design xi Kaiser Filter Parameters Name la Filter type Stop and Cut off Frequencies Wpl 10 25 wst a5 Whe 1 0 Ws2 0 75 Ripple dB PB 420 0 SB 25 0 Close Update Help Warning Applet Window a Kaiser dialog window Script use Name Kaiser Example code lt param name 3 value B3 Kaiser 3 1 gt Equation s Implemented The order and value of B of the Kaiser window are calculated by ag 0 1102 A 8 7 A gt 50 and B 40 5842 A 21 0 07886 A 21 21 lt A lt 50 2 285 AQ 0 perp Aq is the transition band of the filter and A is equal to the smaller of PB and SB M6 6 M6 6 Block name Parks McClellan Notation Parks Mc Description This block designs FIR filters using the Parks McClellan algorithm with min max design Filter specifications are Filter type can be low
108. f the input signal can be examined Pin assignment E Dialog window s e x Plat tools 7 Grid i Graphical zooming Manual adjustment of axes Maan scale linear C dB Displaying the signal as continuous or discrete oT dB linear scale for magnitude plots and deg rad for phase plots e Time domain signals are plotted in terms of time samples 0 0 e Frequency domain signals are plotted in terms of radians 1 0E0 0 100 GridOrn Off Plot cont Apis Auto Graph alues5 tats Close Help Warning Applet Window a Plot dialog window Graphical M2 9 Mame la Arnplitude scale C ilinear dB He Signal Statistics 0 blar Value 16 694 1 Min Value 0 617 2 Signal Lengthisamplesi 100 3 Wean 6 279 4 Variance 12 174 i Standard Deviation 3 4649 5 Total Energy 5160 76 T Power 51 607 E z 10 11 Graphi aluessStats Close Help Java Applet Window Index Value 4 546 9815 4 381 2 14 2 3035 Gare 4047 3 126 10 135 10 498 16 6949 10 043 Amplitude scale linear dB Value 2 511 6 551 10 277 2 4943 12 932 564r 3 476 34r Taaa 4 024 2 505 1 453 Close Help Java Applet window b Plot dialog window Statistics and values Script use Name plot Example code lt param name 3 value B3 plot 3 1 gt Equation s Implemented I lt ly Mean u x n Variance o gt x n u n l n l N 1 N St
109. ference on Acoustics Speech and Signal Processing ICASSP 97 pp U 1175 Munich April 1997 S Ahmadi and A Spanias A New Sinusoidal Phase Modeling Algorithm IEEE International Conference on Acoustics Speech and Signal Processing ICASSP 97 pp I I 1675 Munich April 1997 11 C 39 C 40 C 41 C 42 C 43 C 44 C 45 C 46 C 47 C 48 C 49 C 50 C 52 C 53 C 54 C 55 E Painter and A Spanias A Matlab tool for the evaluation of Speech Coding Algorithms IEEE International Conference on Acoustics Speech and Signal Processing Atlanta May 1996 K Kitsios A Spanias B Welfert and P Loizou An Adaptive Modified Covariance Algorithm for Spectral Analysis IEEE Workshop on Statistical Signal and Array Signal Processing pp 56 59 June 1996 S Ahmadi and A Spanias Low Bit Rate Speech Coding Based on Harmonic Sinusoidal Models International Symposium on Digital Signal Processing pp 165 169 London July 1996 P Loizou A Mekkoth and A S Spanias Telephone Alphabet Recognition for Name Retrieval Applications Proceedings of International Conf on Signal Processing Applications and technology pp 2014 2018 Boston October 1995 G Tucker A S Spanias and P Loizou An HMM based Endpoint Detector for Computer Communication Application Proceedings of International Conf on Signal Processing Applications and technology pp 1969 1973 Boston
110. for Signal Processing Experiments IEEE International Conference on Acoustics Speech and Signal Processing ICASSP 98 DSP 16 Seattle May 1998 A Xavier and A Spanias An Adaptive System Identification Java Simulation for Internet based courseware 17 International Conference on Modeling Identification and Control Grinderwald Feb 1998 G Nair and A Spanias Eigenspace Projections for FIR System Identification 17 International Conference on Modeling Identification and Control Grinderwald Feb 1998 S Ahmadi Andreas S Spanias New Algorithms for Sinusoidal Speech Coding at Low Bit Rates IEEE International Conference on Personal Wireless Communications pp 57 61 December 1997 K Darroudi and A Spanias Robust Speech Coding based on Pole Zero representations and Trellis Coded Quantization The International Conference on Signal Processing Applications amp Technology pp 1709 1713 San Diego September 1997 S Ahmadi and Andreas Spanias A New Phase Model for Sinusoidal Transform Coding of Speec Signals Proceedings of IEEE Mediterranean Conference on Control and Systems CCS Cyprus July 1997 T Painter A Spanias A Review of algorithms for Perceptual Coding of Digital Audio Signals Proceedings of International Conference on Digital Signal Processing DSP pp 179 205 July 1997 M Deisher and A Spanias HMM Based Speech Enhancement using Harmonic Modeling IEEE International Con
111. g File and then Export as Script HTML Export E x Copy and paste this code Applet Code Only ka applet CODE D sp class width 400 height 250 param name num Command walue 2 gt lt l START PARTS gt param name O value BO siggen 1 3 gt param name 1 yvalue B1 filter 2 3 gt param name 2 value B2 plot 5 3 gt param name 3 value B3 coeft 2 gt lt l END PARTS gt l START CONNECTIONS gt param name 4 value C 0 4 1 0 param name 5 yalue C 1 4 2 0 gt param name 6 yalue C 3 3 1 2 gt l END CONNECTIONS gt l START OPEN DIALOGS gt param name z value 0 0 l END OPEN DIALOGS gt l START PART P amp RAMATERS O00 NOT MODIFY gt param name 8 yalue PO7S0 10 0 77 0 0 9 0 00 2 7 4 Triangular Hony param name 9 value P1 param name 10 value P27 c dB Magn caont alze gt param name 11 value P3771 0 1 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 lt l END PARAT PARAMATERS gt el lt r applet Close Help Warning Applet Window Figure 2 J DSP script window Step 3 Using the mouse select the code from the window and press Ctrl C to copy it into the clipboard Some users may use a right click and then Copy depending on the Java version used on their computer Non Windows users should be able to follow a similar procedure Note If
112. g rate conversion and simulate a two band quadrature mirror filter QMF bank The students get familiar with the up sampling and down sampling rules They also study the effects of aliasing and imperfect reconstruction in decimation and interpolation of digital signals Four problems are assigned in this lab In Problem 1 students examine the effect of down sampling and up sampling on FFT spectra In Problem 2 students design a fractional sampler In Problems 3 and 4 students implement and evaluate a two band QME bank and a tree structured QMF using J DSP Lab 6 Introduction to Random Signal Processing This lab is optional and covers elements of spectral analysis of random signals The goal of this exercise is to provide students with the basics of classical and parametric spectral estimation J DSP has functions for estimating periodograms and correlograms It can also estimate parametric autoregressive AR spectra by using the linear predictive coding LPC functions Correlograms are established by connecting the output of the long signal generator to the autocorrelation block This is followed by a connection to a lag window and then to SymCorr and subsequently to the correlogram block The graph panel will then show the correlogram The length of data the window and the length of correlation are selectable and enables students to experiment with trade offs of spectral resolution and statistical variance A task to estimate the spectrum of
113. gital Audio Ted Painter Dept Electr Eng ASU August 2000 T Painter is with Intel Sinusoidal Modeling of Wideband Signals Khosro Daroudi Dept Electr Eng ASU December 1999 K Daroudi is with Intel Adaptive Filters Based on Eigenspace Projections Gopal Nair Dept Electr Eng ASU May 1998 G Nair is with Intel An Improved Approach to Robust Speech Recognition Using Minimum Error Classification Min Tau Lin Dept Electr Eng ASU December 1997 M Lin is now with Solectron in San Jose Low Bit Rate Coding based on the Sinusoidal Model Sassan Ahmadi Dept Electr Eng ASU August 1997 S Ahmadi is with Nokia at San Diego State Based Noise Reduction Using the Sinusoidal Speech Model Mike Deisher Dept Electr Eng ASU May 1996 Mike Deisher is currently with Intel Corporate Research in Portland Robust Speaker Independent Recognition of Alphabet Symbols Philipos Loizou Dept Electr Eng ASU May 1995 Philipos Loizou is now Associate Professor at the University of Texas Dallas 22 9 Single and Multiple Channel Block Adaptive Filters for Active Noise Cancellation Qun Shen Dept Electr Eng ASU Dec 1992 Qun Shen is Ericsson at the Research Triangle Park M S Theses Supervision Completed l 2 10 11 12 13 14 15 16 17 18 19 20 21 22 233 24 MATLAB Implementation of the G 729 V Atti De
114. gnal Processing Spectral Estimation Communications Research Adaptive Filters Speech Analysis Synthesis Coding and Enhancement Voice Processing Algorithms for Multimedia Applications Time Varying Spectral Analysis DSP Architectures Positions Held Aug 1997 present Professor Department of Electrical Engineering Arizona State University Aug 1993 July 1997 Associate Professor Department of Electrical Engineering Arizona State University Aug 1988 July 1993 Assistant Professor Department of Electrical Engineering Arizona State University May 1985 Aug 1988 Graduate Research Assistant Dept of Electrical and Computer Engineering WVU Aug 1983 May 1985 Graduate Teaching Assistant Dept of Electrical and Computer Engineering WVU July 1979 Aug 1981 Tactical Communications Engineer National Guard Honors Awards Memberships 2003 IEEE Fellow 2004 IEEE Distinguished Lecturer in signal processing 2002 IEEE Donald G Fink Prize Paper Award from the IEEE Board of Directors for the paper Perceptual Coding of Digital Audio 2003 Teaching Award for contributions to J DSP IEEE Phoenix Chapter Phoenix January 2003 2002 Researcher of the Year Award IEEE Phoenix Chapter Phoenix January 2002 1997 Award from the Intel Advanced Personal Communications Division Central Logic Engineering Team Recognition Award for outstanding support and leadership of the ASU Team in the Intel GSM Cellular Mobile Telephone Project
115. h 4 6 1992 A S Spanias A Hybrid Transform Method for Speech Analysis and Synthesis Proc EEE International Global Telecommunications Conference GLOBECOM 91 pp 719 724 Phoenix Dec 1991 M Deisher and A S Spanias Adaptive Noise Cancellation Using the Fast Optimal Block Algorithm FOBA Proc IEEE International Symposium on Circuits and Systems ISCAS 91 pp 698 701 Singapore June 1991 P Loizou and A S Spanias Low rate Speech Representation by Vector Quantizing Transform Components Proc IEEE International Symposium on Circuits and Systems ISCAS 91 pp 320 323 Singapore June 1991 P Loizou and A S Spanias Vector Quantization of Principal Spectral Components for Speech Coding at 1200 BPS IEEE Proc International Conference on Acoustics Speech and Signal Processing ICASSP 91 pp 245 248 Toronto May 1991 M Deisher and A S Spanias Real time implementation of a frequency domain adaptive filter on a fixed point signal processor IEEE Proc International Conference on Acoustics Speech and Signal Processing ICASSP 91 pp 2013 2016 Toronto May 1991 P Loizou and A S Spanias Vector Quantization of Principal Spectral Components for Speech Coding at 4800 BPS Presented at the 24Th Asilomar Conference on Circuits Systems and Computers Asilomar Conf Rec Pacific Grove California Nov 1990 A S Spanias A Hybrid Model for Speech Synthesis IEEE International Symposium on C
116. h Enhancement Using the Pseudocepstrum Task 2 A Spanias and K Daroudi 50 pages ASU TRC Technical Report TRC SP ASP 9206 Reported to Motorola Inc November 1992 R 14 Analysis and Implementation of the GSM RPE LTP Algorithm A Spanias P Loizou and G Lim 48 pages ASU TRC Technical Report TRC SP ASP 9205 Reported to Intel Corp October 1992 R 15 Active Noise Cancellation in Ducts 3rd Quarter A Spanias and J Liu 12 pages ASU TRC Technical Report TRC SP ASP 9204 Reported to ANVT September 1992 16 R 16 Speech Enhancement Using the Pseudocepstrum Task 1 A Spanias and K Daroudi 13 pages ASU TRC Technical Report TRC SP ASP 9203 Reported to Motorola Inc August 1992 R 17 Simulation Models for the GSM Channel Coding and Modem Functions A Spanias W Ma and F Tiong 25 pages ASU TRC Technical Report TRC SP ASP 9202 Reported to Intel Corp August 1992 R 18 Fixed Point Implementation of the VSELP algorithm Final Report A Spanias M Deisher P Loizou and G Lim 286 pages ASU TRC Technical Report TRC SP ASP 9201 Reported to Intel Corp July 1992 R 19 Development and Evaluation of Fixed Point Full and Half Rate GSM Coders Progress Report on the Full Rate GSM Speech Codec 3rd Quarter A Spanias P Loizou G Lim and M Deisher 14 pages ASU Report CRR 92070 Reported to Intel Corp June 1992 R 20 Fixed Point Implementation of the VSELP algorithm Task 3 A Spanias M Deisher P L
117. hose that have linear phase A 19 30 improvement is observed here E Pre assessment Assessment of Lab3 B Post assessment O Improvement 120 x OD oO N gt 100 N oO oS AN low O C200 l LO 2 o On N m gt 80 Z CO ies N gt D re O O Y gS 2 t M MN cco 60 LO LO N LO ogo ph NOSKI OES 40 C8 Me FON a Ceol is aN S A gt Ae AN co OO o0 20 8 1 N pE Ee O l HAIE l o AT A U iG EP i Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Question Number FIGURE 4 C THE RESULTS OF LAB 3 ASSESSMENT 11 Questions 8 and 9 are related to finding the group delay from the phase and impulse responses of FIR filters Question 10 1s set to assess whether students are aware of the fact that FIR design by frequency sampling of ideal frequency responses does not yield an ideal filter Lab 4 assessment results Lab 4 is about the FFT Question 1 is related to the symmetry properties of the DFT The improvement observed is 9 47 Question 2 is regarding the resolution of the DFT Question 3 was about the mainlobe and sidelobe characteristics of the rectangular window A 9 23 improvement is observed here This was a relatively long laboratory exercise and our general impression was that the students were able to understand several concepts on the DFT that are not immediately evident from the lecture and the text book The students gained valuable experience on spectral resolution and spectral leakage by
118. ia Cyprus March 12 2002 e General Dynamics Vocoder Technologies for Secure Communications Signal Processing Excellence Series Scottsdale AZ April 19 2002 e Smart Antennas TRC Industry Advisory Board Feb 2001 e An on line signal processing laboratory IEEE DSP Education Workshop Hunt Texas October 15 18 2000 17 Speech Coding University of Manchester April 1999 Speech Processing ASU Systems Science TRC Seminar Tempe February 1996 Sinusoidal Models for Speech Coding Intel Corporate Research Council Portland February 1996 Speech Coding Imperial College London June 1995 Speech Coding Algorithms Texas Instruments Corporate Communications Group Dallas September 1995 On the Utility of the FFT in electromyography ASU ESPE Department Biomechanics Lab March 28 1995 Speech Coding Technologies IEEE Communications and Signal Processing Phoenix Chapter December 6 1994 Several invited conference paper presentations by A S Spanias marked with in the list of conference publications System Identification Algorithms Presentation to the Du Pont Process Control Technologies Panel September 16 1992 Speech Compression Algorithms Presentation to the Connectivity Group at Intel Corporation August 14 1992 Speech Coding Seminar given on February 13 1990 Systems Science and Engineering ASU Speech Coding for PCN Lecture given on March 10 92 Telecommunications Research C
119. ignal Processing COMSOC SP Phoenix Chapter Chair 1993 97 Coordinate local IEEE meetings IEEE Communications and ASSP Society COMSOC SP Phoenix Chapter Vice Chair 1990 93 Membership in National and International Committees Elected Member of the IEEE Signal Processing Society Technical Committee on Statistical Signal and Array Processing formerly Spectrum Estimation and Modeling 1991 1997 Elected Member of the IEEE Signal Processing Conference Board 1993 1999 Elected Member of the IEEE Circuits and Systems Society CAS Technical Committee on Digital Signal Processing 1992 1997 Major Scientific Service in IEEE Signal Processing Society Vice President Conferences IEEE Signal Processing Society 2000 2002 Member Board of Governors IEEE Signal Processing Society 2000 2002 Member Executive Committee IEEE Signal Processing Society 2000 2002 Associate Editor IEEE Signal Processing Letters 2000 2002 Associate Editor IEEE Transactions on Signal Processing 1994 1997 General Conference Co Chair along with D Cochran 1999 IEEE International Conference on Acoustics Speech and Signal Processing ICASSP 99 Phoenix March 1999 Guest Co Editor IEEE_Signal Processing Magazine Jan 2000 Special Issue on Industry Applications Guest Editor EEE Signal Processing Magazine March 2000 Special Issue on Industry DSP Technology Founder and Chair Industry DSP Committee EEE Signal Processing Society Other Conference Activiti
120. ilters J DSP Lab 4 digital oscillators J DSP Lab 2 and QMF filter banks J DSP Lab 6 Several engineering problems are solved within the context of J DSP labs relating to ABET 3e J DSP tasks also satisfy ABET Criterion amp since discrete mathematics and linear algebra are inherent to DSP 7 2 Interactivity The user friendly interface of J DSP allows simulations to be performed interactively The user can examine every signal at any part of the algorithm block diagram By double clicking on a block a dialog window appears that allows interaction and modification of various parameters In EEE 407 J DSP enables the user to go through a variety of laboratory exercises that cover several aspects of DSP theory With regard to two way communication it is possible for the instructor to develop web and streaming content where feedback on J DSP tasks is given automatically Although we have in place the J DSP capabilities for scripting that allow integration of J DSP demos with web content and multimedia presentations we have not yet finalized a formal web class We are currently in the midst of producing a DSP web class with streaming content that will incorporate interactive sessions with J DSP The goal is to eventually produce an environment similar to that shown in Appendix A Fig A 5 7 3 Application of Concepts to other areas When students complete the J DSP lab sequence they are required to propose and complete a project that relates to
121. iption Parameters List Lists the input parameter values SNR Calculates the signal to noise ratio between two signals Statistics Calculates signal statistics of the input signal Window Windows a time domain signal Mixer Adds subtracts two signals D Sampling Down samples a signal U Sampling Up samples a signal Convolution Performs convolution of two input signals Parameters List SNR statistics Window Adder D Sampting U Sampling Convolution M3 1 Block name Parameters list Notation List Description This block tabulates the signal values applied at its input in a text box No action is taken on the signals that are passed directly to the outputs Typical signal types allowed are filter coefficients tme domain and frequency domain signak Pin assignment rs B Coefficients b Q 1 0 b1 0 89999 b 2 0 0747 b 3 0 42643 b 4 0 61349 b 5 0 01214 Input signal 0 38561 bl 0 27917 Coeffilenj ja Coefficients Siglenj 256 Close Help Java Applet Window a List dialog window Script use Name list Example code lt param name 3 value B3 list 3 1 gt M3 2 M3 2 Block name Signalto noise ratio Notation SNR Description This block calculates the signakto noise ratio SNR value in dB between two signals The reference signal is given as input to the upper input pin Pin assignment Reference signal Proc
122. ircuits and Systems ISCAS 90 Conf Proc ISCAS 90 Vol 2 pp 1521 1524 New Orleans May 1990 Presentations by Andreas Spanias 13 C 71 C72 C 73 C 74 C 75 C 76 C77 C 78 C 79 C 80 C 82 C 83 C 84 C 85 A S Spanias S B Jonsson and S D Stearns Transform Coding Algorithms for Seismic Data Compression IEEE International Symposium on Circuits and Systems ISCAS 90 Conf Proc ISCAS 90 Vol 2 pp 1573 1576 New Orleans May 1990 W B Mikhael and A S Spanias A Least Squares Pole Zero Algorithm in the frequency and Walsh Domains with applications to speech representation IEEE International Symposium on Circuits and Systems ISCAS 90 Conf Proc ISCAS 90 Vol 2 pp 1331 1334 New Orleans May 1990 W B Mikhael and A S Spanias Direct Coding of a Class of Non Stationary Signals Based on Mixed Transforms IEEE International Symposium on Circuits and Systems ISCAS 89 Conf Proc ISCAS 89 Vol 1 pp 280 283 Portland May 1989 W B Mikhael and A S Spanias Fourier Walsh Representation of a Class of Non Stationary Signals IEEE International Symposium on Circuits and Systems ISCAS 89 Conf Proc ISCAS 89 Vol 3 pp 1768 1771 Portland May 1989 W B Mikhael and A S Spanias Reduced Bit rate Representation of Speech Using Mixed Fourier Walsh Transforms 22nd Asilomar Conference on Circuits Systems and Computers Pacific Grove California pp 366 370
123. is given by b b ue n x by n x n b n x n 1 where b is the filter coefficient vector x is the input vector and by n x n WN 1 N 1 e n d n yb n x n 1 is the error signal The step size u is the adaptation constant that l 0 controls the rate of convergence M6 8 M6 8 Block name Frequency sampling Notation FregSamp Description This block designs a linear phase finite impulse response FIR filter based on the frequency sampling method In he frequency sampling method an FIR impulse response is obtained by applying an IFFT on samples of a desired frequency response The desired frequency response 1s drawn using the dialog window shown below Pin assignment B T Filter coefficients Freg Samp Dialog window s User entry 2 Frequency Sampling Block Ed Number of line segments used to draw the desired freq User entry 1 The number of samples used in the Line Segments 1 Samples BR re art an will export an equivalent number of FIR coefficients response User entry 3 Consecutive placement of points on the drawing area creates line segments related with the desired freq response Auxiliary lines are drawn automatically to assist the user to visualize the resulting frequency response Close f Java Applet Window a FreqSamp dialog window Script use Name FreqSamp Example code lt param name 3 value BO FreqSamp 1 7 gt Equation s
124. ishing an assigned laboratory exercise with J DSP Lab 2 The Z transform and Frequency responses Q1 The contents of this exercise improved your understanding of the concepts of the Z transform S La Agree s Neutral E S Strongly Agree Disagree Strongly Disagree Q2 Performing this exercise you learned how to generate a sinusoid with a digital filter Strongly Agree Agree Neutral Disagree ON fi N Strongly Disagree Q3 Can you now understand more clearly the relationship of the impulse response with the transfer function s Yes s No Q4 Did you experience any problems in terms of connection time to implement etc Please describe Q5 Was there enough information in the help screens to assist you in using the blocks La Yes L No La Did not use the help screens Q6 Is there any particular concept which you could not grasp earlier from your text book class but became clear from this exercise If so please describe El Q7 Performing the exercises you are now more comfortable with these topics s Yes s No Q8 Setting up the required lab simulations was pretty easy L Strongly Agree Agree Neutral Disagree oo n Strongly Disagree Q9 Can you suggest an exercise along the lines of this one Q10 Please suggest possible improvements relative to this lab such as redesigning of dialog box of a block what to add or delete etc E Lab 3 Frequency responses and pole zero
125. ith Phylon in San Jose CA High Sample Rate Architectures for the BLMS Algorithm Karkada Srikanth Dept Electr Eng ASU Dec 1993 Karkada is currently with Phillips in San Jose CA Speech Processing Using the Bispectrum Ralph Fulchiero Dept Electr Eng ASU Aug 1993 Ralph is currently with Motorola GSTG Frequency Domain All Pole Spectral Matching With Applications to Speech Gim Lim Dept Electr Eng ASU Dec 1992 Gim 1s currently with Intel Speech Analysis and Enhancement Using Higher Order Statistics Ines Jebali Gdoura Dept Electr Eng ASU May 1992 Ines is with the Telecom company in Tunisia Digital Image Restoration using the Pseudo Cepstrum Ye Quang Chen Dept Electr Eng ASU Dec 1991 Ye Quang is with the Department of Defense in Taiwan The Discrete Wavelet Transform and its Application To Signal Reconstruction Gen Fuh Hu Dept Electr Eng ASU Dec 1991 Gen Fuh 1s in Taiwan Experimental Analysis of Frequency domain Adaptive Noise Cancellers M E Deisher Dept Electr Eng ASU May 1991 Mike completed a Ph D at ASU is now with the Corporate Research labs at Intel Low rate Speech Representations by Vector Quantizing Transform Components P Loizou Dept Electr Eng ASU May 1991 Design Considerations for a 94 GHz Pulsed Doppler Radar System with Interactive Computer Processing J W Nehrbass Dept Electr Eng ASU May 1991 Transform Coding of Seismic Data S
126. ive effort with a large volume of supporting text manuals assessment and dissemination materials We will be grateful if the reviewers in their evaluation take the time to look at our submitted Appendices that demonstrate well our commitment to this educational task To our knowledge we are the first to submit assessment of labware to the OERL database sponsored by NSF 7 1 Instructional Design 7 1 1 Learning Objectives Our learning objectives have been stated in Section 3 of this submission packet The assessment of our objectives has been summarized in Section 5 and is detailed further in Appendix C The students perform a sequence of laboratory exercises Appendix D using the J DSP software The exercises have been designed carefully to support the learning of the topics covered in the DSP class that range from discrete time filter basics to filter design and spectral analysis By providing hands on experiences on the class topics J DSP enables the student to associate abstract mathematical concepts of DSP theory to actual implementation issues 7 1 2 Support and Measurement of Learning Objectives Because the laboratory assignments are well coordinated with in class EEE 407 lectures that are available to students either real time or through synchronous and asynchronous webcasting the students are well aware of the learning objectives for each topic From assessments Appendix C it is evident that J DSP enabled the students to comprehend
127. l X No 2 April June 2000 en The DSP prototype was originally funded by Arizona State funds and subsequently extensions to this prototype have been funded by NSF The functionality ofthis software has been extended to cover other topics such as communications and controls In addition the infrastructure of the software environment was enhanced to provide additional features that made it very useful to DSP instructors The software was formally disseminated at the 2002 IEEE DSP workshop and at the 08t eaye 0p2 E12 NY30 JHL 40 391440 SNINBENIDNS OSn wory Weg Bg e00z o1 unp Page Two NEEDS Award Committee June 9 2003 FIE conference i in Boston i in the fall of 2002 Pee sees 50 university professors obtained evaluation copies from universities including MIT Georgia _ Tech Rice University U Minnesota and UPenn In all there are about fifty worldwide beta sites that are being established including some in major _ universities in Argentina China Finland Puerto Rico iii Sweden Switzerland Turkey and the United Kingdom Functionality extensions have been reported by Prof Spanias and his team at _ ICASSP 2001 A major capability was developed for this tool by designing a script interpreter that enables users to embed interactive J DSP simulations in mmi content This particular capability enables instructors to design web courses with interactive visualization content that is developed using J DSP The value of this featu
128. l independent study course entitled Signal Processing Using Higher Order Statistics EEE 790 four Ph D students Spring 92 This group met once a week for three hours and material was presented from several research papers Andreas Spanias introduced the subject during several lecture sessions and students took turns presenting the results of research papers in Higher Order Statistics Developed and supervised an advanced level independent study course entitled Speech Processing EEE 790 one Ph D student Fall 92 The student studied voice recognition algorithms developed software in MATLAB for one of the algorithms and presented his results in a report Developed and supervised a graduate independent study course entitled Speech Coding for Multimedia Applications EEE 590 one M S student Spring 1994 The student studied speech coding algorithms developed software in MATLAB for one of the algorithms and presented results in a report Developed MATLAB Educational Software for Speech Coding sample on URL http www eas asu edu trcsip painter educsw html Also described in a publication in the IEEE Trans On Education An Educational Software Tool for the Study of Speech Coding Algorithms in a DSP Class Andreas Spanias and Ted Painter Special Issue on DSP Education EEE Trans on Education Vol 39 No 2 pp 143 152 May 1996 Developed teaching material consisting of 400 viewgraphs for the DSP course EEE 407 This w
129. l length of the signal Now take a look at the output signal In the time domain it is very hard to see that a sinusoid is present However if you view the signal in the frequency domain with an FFT size of 256 then you still find a peak at approximately 0 31 Step 3 4 Change the amplitude of the sinusoid up or down and observe the spectra FFT plot Try different values to make the sinusoid to dominate the noise signal or be masked by the noise signal Remember the movie The Hunt for Red October which was about a stealth Soviet sub marine defecting In that movie they showed sonar operators viewing FFT spectra and listening to sonar signals as they were searching for submarine propeller signatures quasi periodic signals in ocean noise random broadband signals Stealth submarines have among other things weak broadband propeller signatures that can be masked easily by ocean noise M1 8 J DSP Editor MANUAL IZONA STATE J DSP Editor concept by Prof A Spanias UNIVERSITY Copyright 1997 2003 Arizona Board of Regents Andreas Spanias Biography Andreas Spanias is Professor in the Department of Electrical Engineering at Arizona State University ASU His research interests are in the areas of adaptive signal processing and speech processing While at ASU he has developed and taught courses in DSP adaptive signal processing and speech coding He has also developed and taught continuing education short courses in digi
130. lation Tool For Introducing Algebraic CELP ACELP Coding Concepts In A DSP Course EEE 2002 DSP Workshop Callaway Georgia October 2002 A S Spanias V Atti Y Ko T Thrasyvoulou M Yasin M Zaman T Duman L Karam A Papandreou K Tsakalis On Line Laboratories For Speech And Image Processing and for Communication Systems Using J DSP IEEE 2002 DSP Workshop Callaway Georgia October 2002 R Ramapriya and A Spanias A Simulation Tool for introducing MPEG Audio MP3 concepts in a DSP course To appear in Proc IEEE International Conference on Acoustic Speech and Signal Processing ICASSP 2002 Orlando May 2002 S Bellofiore J Foutz C Balanis A S Spanias T Duman Signal Processing and Communications Algorithms for Array Antennas Proc IEEE International Symposium on Circuits and Systems ISCAS 02 Phoenix May 2002 J Foutz and A Spanias Adaptive Eigen Projection Algorithms for 1 D And 2 D Antenna Arrays Proc IEEE International Symposium on Circuits and Systems ISCAS 02 pp 201 204 Phoenix May 2002 T Painter and A S Spanias Sinusoidal Analysis Synthesis of Audio using Perceptual Criteria Proc IEEE International Symposium on Circuits and Systems ISCAS 02 Vol 2 pp 177 180 Phoenix May 2002 Bellofiore S Balanis C A Foutz J Spanias A Impact of smart antenna designs on network capacity IEEE Antennas and Propagation Society International Symposium
131. lem 6 students are asked to compare the sidelobe levels obtained for the filters designed using the Parks McClellan method the Kaiser design and the frequency sampling method Problem 7 deals with the design of IIR filters using bi linear analog filter approximations In particular students are asked to design and compare Butterworth Chebychev I Chebychev II and Elliptic digital filters Lab 4 The Fast Fourier Transform FFT In this lab students learn various concepts related to the use of the DFT and the FFT In particular in this lab students gain familiarity with the estimation of DFT spectra DFT spectral leakage DFT resolution the Parseval s theorem for the DFT FFT properties and symmetries and signal estimation and reconstruction using the FFT In Problem 1 students examine symmetries of the FFT In Problem 2 students observe the effect of zero padding and windowing on the FFT spectra The blocks used in J DSP for Problem 2 are shown in Figure 3 In Problem 3 the students are asked to examine and compare the FFT spectra of various signals In Problem 4 students are provided with two sinusoids that are closely spaced in the frequency and are asked to examine the FFT spectra with several windows Sig Ger Widow FFT Plot FIGURE 3 BLOCKS USED IN FIR FILTER DESIGN BY WINDOWING IN LAB4 PROB 2 Lab 5 Multi rate Signal Processing and QMF banks The goal of this exercise 1s to examine the effects and the use of the samplin
132. log box of a block what to add or delete etc Lab 5 The Fast Fourier Transform FFT Q1 The contents of this exercise helped you understand the general concepts of using Fast Fourier transform in signal analysis L Strongly Agree Agree Neutral Disagree annn nA Strongly Disagree Q2 Which part of the exercise helped you the most to understand how and when to use the FFT E Q3 The exercise helped you to clearly visualize signal symmetries on the FFT spectra Strongly Agree Agree Neutral Disagree nnn fi nA Strongly Disagree Q4 This lab strengthened your perception about various window trade offs Strongly Agree Agree Neutral Disagree Hho fe Strongly Disagree Q5 By performing this exercise you understand that spectral resolution of the FFT is limited by frame size window type and window size s Yes s No Q6 Did you experience any problems in terms of connection time to implement etc Please describe E Q7 Was there enough information in the help screens to assist you in using the blocks La Yes L No La Did not use the help screens Q8 Is there any particular concept which you could not grasp earlier from your text book class but became clear from this exercise If so please describe Q9 Performing the exercises you are now more comfortable with these topics L Yes L No Q10 Setting up the required lab simulations was pretty easy L Strongly Agree Agree Ne
133. lt font size 2 gt lt br gt All material Copyright ttc 1997 2002 Arizona Boar Department of Electrical Engineering lt br gt Multidisciplinary Initiative on Distance Learning ASU lt br gt ogo E PR K Line 160 Column 6 EJ Z5 seconds over 28 8 A Figure 3 Pasting a J DSP script using Ms FrontPage lt applet gt Step 4 2 Saving into a new HTML file a Use any text HTML editor to create a new file b Copy and paste the entire HTML code from the script window into the text HTML editor c Save the file with an extension of htm or html so that any browser can recognize it For example you can name the file myjdsp html Important note In every case make sure that the HTML file containing the script is saved where the J DSP editor class files are saved otherwise the script will not run M10 3 Step 5 After saving the HTML file in the same directory as the J DSP script files load the file in a browser and press the Start button This should start the J DSP Editor and load the saved simulation Make sure you place a link to this HTML file from your web page so that others can have access to it You have now created a J DSP script If you are interested to learn more on how J DSP scripts work you can read further to section 3 otherwise you can safely stop here 3 Generating scripts manually Writing the code Typically only a few lines of code are necessary to set up and execute a simulation The f
134. ltering 14 18 system identification based on ellipsoidal bounded error constraints 13 e He developed a new concept for on line DSP laboratories using a novel Java object oriented programming environment called Java DSP This concept was adopted by ten university and industry partners and was recently awarded a sizeable three year NSF grant for disseminating and extending J DSP to communications and other areas e He demonstrated outstanding and sustained leadership in academic industry and IEEE activities as evidenced by various project chip development and EEE paper awards as well as devoted service in upper level IEEE volunteer activities Select Papers Published 1 T Painter and A S Spanias Perceptual coding of digital audio Proc IEEE vol 88 no 4 pp 451 513 Apr 2000 This 63 page award wining paper was the main article featured in the April 2000 issue of the Proc of the IEEE It describes the theory and research on MPEG audio and cinema algorithms and has more than 400 references This paper won the prestigious 2002 IEEE Donald G Fink Prize Paper Award for best survey paper across all IEEE societies and publications A Spanias principal investigator and Ph D advisor 2 A S Spanias Speech coding a tutorial review Proc of the IEEE vol 82 no 10 pp 1441 1582 Oct 1994 A Spanias principal investigator 3 M Deisher P Loizou G Lim A S Spanias and B Mears A new highly integrated
135. ly graphical user interface GUI that is freely and universally accessible from the Internet using a browser such as the MS Internet Explorer The J DSP GUI enables visual programming of DSP tasks on any platform directly from the browser In Appendix A we show several figures that present the GUI of the software We also encourage the NEEDS reviewers to visit the web site http jdsp asu edu and or use the enclosed CD ROM and view the AVI files that demonstrate some key functions of J DSP J DSP is freely accessible platform independent and created for non for profit use J DSP enables students to establish and run quick simulations on the web students to see on line interactive demos embedded in web lectures students to perform on line computer laboratories on the web instructors to assign J DSP based exercises instructors to embed interactive demos in their web lecture content The J DSP Version 1 CD ROM ISBN 0 9724984 0 0 is approximately 42 000 lines of Java code To fully support J DSP Prof Spanias has developed a series of J DSP laboratory exercises see Appendix D that have been used in EEE 407 Assessment of J DSP is carried on a semester by semester basis and web based assessment instruments have been developed and posted on the J DSP web site A more formal and detailed assessment of the J DSP software and its associated EEE 407 lab exercises was conducted recently and results will be presented at the 2003 IEEE A
136. m window length is 256 samples Pin assignment sf Dialog window s Vai india Hamming L 40 Led Er aL L Samples Type Hamming f Length jao Beta Kaizer 3 0 Close Update Help Warning Applet window a Window dialog window Script use Name window Example code lt param name 3 value B3 window 3 1 gt Equation s Implemented y n w n x n x n input signal w n windowing function y n windowed signal M3 5 M3 5 Block name Mixer or Adder Notation Mixer Description Adds or subtracts two signals Pin assignment 2 Input signal x n SSS 3 Output signaly oo 21 ia PT 6l S Dialog window s e ef ADLI iad Pp F c amelie 2 OME ADLE ee Close Update Help warning Applet Window a Mixer dialog window Script use Name mixer Example code lt param name 3 value B3 mixer 3 1 gt Equation s Implemented y n x n x n x n input signalat pin 1 X2 n input signal at pin 2 y n output signal M3 6 M3 6 Block name Down sampling Notation D Sampling Description Down samples the input signal by an integer factor M Pin assignment Input signal Down sample d signal CCC E Slo Sl Dialog window s Down Sampling a x Down 5ampling Down Sampling Rate 1 10 fi Close Update Help warning Applet window a D Sampling dialog window Script
137. ment for students and instructors This java based software provides a visual simulation environment that is accessible on the internet through a simple browser the original idea with J DSP has been to enable on line computer laboratories Professor Andreas Spanias has done an excellent job incorporating J DSP into his undergraduate curriculum since J DSP is being used routinely now in the undergraduate DSP course at Arizona State University with an on line laboratory manual with exercises Instructors can built Java based DSP simulations and embed them in their courseware Articles describing J DSP have been published in conferences and journals Several papers have been presented in engineering education conferences and special sessions These papers describe new functionality in developed for J DSP as well as assessment results obtained from using the software in on line laboratories delivered in distance learning classes In the DSP workshop a paper on functionality extensions that support communications controls speech processing and image processing has been presented o __J DSP is indeed a unique educational software tool that will impact many of the courses related to linear systems and signal processing We enthusiastically support the nomination of this unique on line laboratory software for the NEEDS award ae ae Sincerely me i o ey en S COL p ca Dr Paul Hasler Associate Professor a E Department of Electrical
138. n CRP 91254 Jun 1991 May 1993 60 000 00 21 CO PI A S Spanias with D Morrell Voice and Data Processing Algorithms for the NLOS Communications McDonnell Douglas Helicopter CRP 91203 Mar 1991 Mar 1992 20 332 00 CO PI A Spanias with D Morrell and D Cochran Radar IR Image Object Classification Motorola GEG CRP 90255 Jan 1990 Jan 1991 49 916 00 PI A S Spanias Spectral Analysis of EMG Agent Institutional Biomedical Research Support Program ASU CRP 91081 Jan 1991 Dec 1991 8 500 00 PI A S Spanias A Hybrid Model for Speech Coding Department of Defense DOD CRP 90036 Feb 1990 Jan 1993 225 290 00 PI A S Spanias Low Rate Speech Coding Using Complex Spectral Functions US West Advanced Technologies CRP 89164 June 1989 May 1990 40 395 00 Consulting Intel Corporation Architecture Design Speech Coding Algorithms Inter Tel Communications Worked on Adaptive Echo Cancellers for the Athena Project Motorola Inc Worked on Speech Recognition Algorithms 1992 Texas Instruments DSP Training Sept 1995 The Cyprus Institute of Neurology and Genetics Participated in Spectral Analysis of Electromyographic Signals for Automatic Diagnosis of Neuromuscular diseases Student Theses and Dissertations Supervised Ph D Dissertation Supervision Completed L 2 Adaptive Algorithms for GPS systems B Badke Dept Electr Eng ASU Dec 2002 Perceptual Coding of Di
139. nal processing and communications systems as educators and engineers According to J DSP assessment results which will appear in the IEBE ASEE FIE 03 proceedings students reported that it took most of them less than fifteen minutes to become familiar with running simulations on J DSP This was without use of a manual and by just using a small introduction in their first DSP exercise on the web In the same paper pre and post assessment results revealed that the J DSP tool helped students understand several concepts that were not immediately evident from their text book or class lectures In fact the study showed that in some cases learning was attributed exclusively to using J DSP There are a total of seven papers that will be presented in this year s FTE 03 highlighting different aspects of the J DSP software including extensions in communications controls and image processing and recently functions that can be used to introduce high school students to the processing and interpretations of everyday signals The papers are co authored by Andreas Spanias and his collaborators and are listed below e A Control Systems Simulation Environment For Distance Learning Labs And Assessment 33rd ASEE IEEE FIE Conference Boulder CO Nov 5 8 2003 e Advanced Concepts in Time Frequency Signal Processing made Simple 33rd ASEEAEEE FIE 03 Boulder November 2003 gt e Assessment of the Java Dsp J DSP On Line Laboratory Software
140. ncryption Algorithms for the Phoenix Architecture Sponsor Intel Corp Amount 192 781 00 CRP 92373 DWT 4473 Aug 1992 Dec 1993 PI A S Spanias CO PI Jennie Si Performance Evaluation of Voice Recognition Algorithms Sponsor Motorola Inc 19 845 00 February 1993 July 1993 PI A S Spanias Enhancement of Speech Using the Pseudocepstrum Sponsor Motorola GEG 39 955 00 CRP 92265 DWT 4460 February 1992 February 1993 PI A S Spanias Development and Evaluation of Fixed Point Full and Half Rate GSM Coders Sponsor Intel Corp Amount 233 463 00 CRP 92079 DWT 4432 Date September 1991 December 1992 PI A S Spanias Active Noise Cancellation in Ducts Sponsor Active Noise and Vibration Technologies Amount 27 682 00 CRP 92039 DWT 8504 Date August 1991 December 1992 PI A S Spanias Fixed Point Implementation of the VSELP algorithm Sponsor Intel Corp Amount 55 984 00 CRP 91289 DWT 4423 Date May 1991 June 1992 PI A S Spanias Transform Coding for Seismic Data Compression Sponsor Sandia National Laboratories SNL CRP 90009 19 982 00 DWJ 6150 November 1989 October 1990 PI A S Spanias and overall project director and 13 other CO PIs from four different colleges CEAS CLAS COE and CEE Multidisciplinary Research on Multimedia Technologies for Distributed Learning Using the Intel PC and the Internet 67 000 Intel Corporation PI D Evans CO PI A S Spanias and 9 other CO PIs A
141. nd evaluate the level of student s understanding of the key DSP concepts before and after performing a particular J DSP lab assignment Thus the statistics obtained from these assessments give us feedback on how the J DSP lab assignments helped the students in learning the key concepts on a particular topic The pre post quiz and the lab are assigned after the relevant theory has been introduced in class We did this in order to ensure that all the students have had some or ideally the same exposure to the topics covered in the lab so that we can isolate specifically the effect of J DSP labs in their learning The students are asked to complete the pre lab assessment before working on that lab After they complete and submit the lab assignments they complete the post lab assessment The questions on the post lab assessments are same as the pre lab assessments but given in a different order 5 1 Results of Pre post Lab Assessments Pre post assessment results are shown graphically in Figures 4 and 5 for Labs 1 through 5 Lab 6 was optional and by the time this report was submitted we did not have the data available The assessment results of each of these labs are discussed below Lab 1 assessment results From Figure 6 we see that a total of 6 questions are assigned in Lab 1 assessment Lab 1 is related to the Z Transform and the frequency response In Question 1 the students are asked to determine the impulse response for a given transfer fun
142. ndow a Coefficient dialog window by line and tabular Script use Name Coeff Example code lt param name 3 value B3 coeff 3 1 gt Equation s Implemented y n bix n i ayn i x n input signal y n output signal a feedback coefficients b feed forward coefficients M2 5 M2 4 Block name Junction Notation Junction Description This block propagates its input signal at its two outputs The input signal can be either time domain frequency domain or filter coefficients The Junction block essentially allows other blocks to share the same signal or parameters Pin assignment aT Dialog window s None Script use Name junction Example code lt param name 3 value B3 junction 3 1 gt Equation s Implemented x n y n z n x n input signal y n output signal at first output pin z n output signal at second output pin M2 6 M2 5 Block name Filter Notation Filter Description This block filters the input signal based on the provided numerator and denominator coefficients and the standard difference equation The filter coefficients must be provided using the Coeff block An option is provided to start with zero initial conditions or non zero initial conditions Pin assignment Input signal x n Filter coefficients Filtered signal y n Feedback and feed forward coefficients a and b eo Dialog window s Filter Bl
143. ns Selected Areas in Communications and Geoscience and Remote Sensing Also reviewed papers for IEE Proceedings Part I ACM the Journal of Adaptive Signal Processing and Control the 1991 1992 and 1993 International Phoenix Conference on Computers and Communications IPCCC 91 IPCCC 92 IPCCC 93 the IEEE Workshop on Statistical Signal and Array Processing the 1993 1994 1995 1996 1999 2000 2001 IEEE International Conference on Acoustics Speech and Signal Processing the 1996 IEEE International Symposium on Circuits and Systems Reviewed Books Digital Signal Processing by Mitra 1996 Papoulis 2nd Edition Signals and Linear Systems book for McGraw Hill 1995 Higher Order Statistics for Prentice Hall in 1993 Reviewed NSF Proposals Short Courses and Tutorials Developed and taught short courses devoted to continuing engineering education in the areas of Digital and Adaptive Signal Processing and Speech Coding These are Fundamentals of Digital Signal Analysis This three day lecture and computer laboratory course was given on July 31st 1990 at Makarios Hospital Nicosia Cyprus January 7th 1991 at ASU September 16 1991 at ASU November 4 1992 at Motorola GEG April 23 1993 at ASU March 14 16 1994 at ASU September 1995 Texas Instruments June 1996 1997 1998 1999 2000 Phoenix Support material 400 pages Fundamentals of Adaptive Signal Processing This course 8 hours was sponsored by the Indus
144. nt 3 1 gt Equation s Implemented u law can be stated as losd Hin 0 logd p A law can be stated as __4 xnl 0 x 1 Aand _ LECA tap I A x 1 14 logA 1 log A where Xin and Xou are the normalized input and output signal amplitudes and A 7 M2 14 Section M1 Introduction 1 1 General Information on J DSP J DSP is an object oriented Java tool where J DSP stands for Java Digital Signal Processing J DSP has been developed at Arizona State University ASU and is written as a platform independent Java applet that resides either on a server or on a local hard drive It is accessible through the use of a web browser J DSP has a rich suite of signal processing functions that facilitate interactive on line simulations of modern statistical signal and spectral analysis algorithms filter design tools QMF banks and state of the art vocoders All functions in J DSP appear as graphical blocks that are divided into groups according to their functionality Selecting and establishing individual blocks is done using a drag and drop process Each block is linked to a signal processing function Fig 1 shows the J DSP editor environment and Fig 2 shows details on the drop down menus and the signal processing functions of J DSP A simulation can be started by connecting appropriate blocks from left to right Signals at any point of a simulation can be analyzed and plotted through the u
145. nual it is possible to learn the basics of DSP by using J DSP s Yes s No Q10 Should the J DSP editor be established as a full fledged tool s Yes s No Part 3 of 3 Q1 What would you change in the J DSP program to make it more useful and user friendly gi Q2 Please give your views for improvement of the help screens foal Q3 Was the process of entering filter coefficients intuitive Any suggestions Q4 Do you suggest any other signal types for the Sig Gen block foal Q5 Please describe any errors or bugs you encountered og ae Ie LAB SPECIFIC ASSESSMENT CONTENTS MUOU CON ements rt erg eee eer ar ent Pr a er ec re ery ne reentry eee sere oor ere ee 2 DS PAD AS SiS ea EEE E E E E E EE 3 Desc Donor Me Labs e a a cet Mea and tetas 4 aD UID MT On TOC OCUN saa r a a a a a a a a a arate 7 PEEP OSU Isa Asse Ss meN S ars ca et esterases T TE anaes cape TE 8 5 1 Results of Pre Post Lab Assessments cccccsssscccesscccesscccesecceansccceessscneesscesessseceesesenensseeeens 8 General ASSESSING T ennaii e a thdesaivhidenaen Mitac beaimedeh nena ain 13 COn ON re a E E E T E A ie eis 17 ASSESSMENT OF THE JAVA DSP J DSP ON LINE LABORATORY SOFTWARE This report presents assessment results of the Java DSP J DSP on line laboratory J DSP software has been developed from the ground up at Arizona State University ASU to support the computer lab portion of the senior level DSP course EEE407 The software
146. o view the LP coefficients and reflection coefficients Pin assignment 2 gt Description LP coefficients of order 10 a TAE Reflection coefficients k IE or gt Dialog window s Reflection Coefficients Order ho Refl Coefficients RC LPC KO 0 2946 ki 0 0402 Ke 0 0976 Kap 0 0172 k4 0 2431 KSE 0 141 KIS 0 2003 kI 0 1485 kK 0 0062 KIST 0 016 Close Help Java Applet Window a LPC gt RC dialog window Script use Name Ipce2re Example code lt param name 3 value B3 Ipc2re 3 1 gt M 8 3 M8 5 Block name RC to LPC Notation RC gt LPC Description This block computes the LP coefficients a from the reflection coefficients k Pin assignment Reflection coefficients k LP coefficients of order 10 a E 6 Dialog window s LP Coefficients Order 10 LP Coefficients a0 7 0 amy 0 2733 aja 0 0063 apy 0 0265 at 0 0878 aji 0 1899 als 0 0832 afm 0 2359 afs 0 1503 aj 0 0106 ajio 0 016 Close Update Help Java Applet Window a RC gt LPC dialog window Script use Name rc2lpc Example code lt param name 3 value B3 re2Ipc 3 1 gt Ms8 4 M8 6 Block name RC to LAR Notation RC gt LAR Description This block converts the reflection coefficients to log area ratios LARs Pin assignment Reflection coefficients k l Log area ratios LARs Dialog window s Log Area Ratio
147. ock y xj Filtering Filter Initial Conditions IC C Fiter with non zero IC Close Help Warning Applet Window a Filter dialog window Script use Name filter Example code lt param name 3 value B3 filter 3 1 gt Equation s Implemented L M y n bix n i ayn i i 0 i l x n input signal y n output signal a feedback coefficients b feed forward coefficients M2 7 M2 6 Block name Frequency response Notation Freq Resp Description This block calculates and displays the frequency response of a filter It can be connected to any block that can generate filter coefficients In its dialog window the top plot displays the magnitude in dB or linear scale and the bottom plot shows the phase Pin assignment Description Feedback and feed forward coefficients a and b Dialog window s Java Applet Window a Frequency Response dialog window Script use Name freqresp Example code lt param name 3 value B3 freqresp 3 1 gt Equation s Implemented J be l H e a feedback coefficients b feed forward coefficients M2 8 M2 7 Block name Plot Notation Plot Description This block primarily plots the signal at its input in an x y axis coordinate system It can also display values in text form and calculate some basic signal statistics The magnitude magnitude squared real part imaginary part and phase o
148. oftware has been thoroughly verified and formal dissemination occurred at the 2002 FIE Conference and at the 2002 IEEE DSP Education workshop in Atlanta The software has been disseminated to more than 50 instructors throughout the world Although initially J DSP assessment was carried within the course evaluation forms in the last year we assessed separately the degree of learning attributed specifically to J DSP We focused in particular on assessing whether J DSP accelerated the learning curve in the DSP class Both on line and off line materials have been developed Two types of on line forms have been developed i e general and concept specific In the general forms the students are asked to provide general qualitative and quantitative evaluation of the J DSP concept including logistics accessibility convenience demographics academic standing etc In concept specific forms students evaluated each laboratory task with regard to its impact on learning specific DSP concepts Our newest assessment instruments assess learning attributed specifically to J DSP by testing DSP concepts before pre assessment the J DSP lab and after post assessment the J DSP lab In this report we present detailed qualitative and statistical results of this comprehensive assessment effort and describe how J DSP contributes to the learning of several key DSP concepts 2 DSP LAB ASSIGNMENTS The laboratory assignments of the EEE407 DSP course are designed to
149. oftware works 5 J DSP Labware Assessment J DSP assessment was carried through the years as part of the evaluation of the EEE 407 class Specific evaluation for J DSP was also carried including evaluations by web site visitors required student evaluations of each lab after EEE407 students submit their reports and specific differential evaluations of each J DSP lab with pre and post lab tests that measure the degree of learning specifically attributed to the usage of J DSP We include the following documents in the NEEDS package e An assessment report is included in Appendix C 1 e Sample assessment forms are shown at the end of Appendix C 1 e Specific comments by EEE 407 students are on Appendix C 2 e Lists of faculty that received J DSP and agreed to form beta sites are included in Appendix C 3 A copy of the dissemination package that was distributed freely at the 2002 IEEE DSP Workshop in Atlanta and IEEE ASEE FIE 02 in Boston is also included Reference letters of faculty and industry engineers that agreed to evaluate the software are in Appendix B Although initially J DSP assessment was carried using the course evaluation forms during the last academic year we assessed separately the degree of learning attributed specifically to J DSP We focused in particular on assessing whether J DSP accelerated the learning process in the DSP class Both on line and off line instruments have been developed Two types of on line forms have been devel
150. oizou and G Lim 30 pages ASU Report CRR 92069 Reported to Intel Corp June 1992 R 21 Active Noise Cancellation in Ducts 2nd Quarter A Spanias and J Liu 20 pages ASU Report CRR 92066 Reported to ANVT June 1992 R 22 Development and Evaluation of Fixed Point Full and Half Rate GSM Coders Progress Report on the Full Rate GSM Speech Codec Ist amp he Quarter A Spanias P Loizou and G Lim 19 pages ASU Report CRR 92057 Reported to Intel Corp March 1992 R 23 Fixed Point Implementation of the VSELP algorithm Phase 2 Tasks 1 and 2 A Spanias M Deisher P Loizou and G Lim 18 pages ASU Report CRR 92045 Reported to Intel Corp March 1992 R 24 Active Noise Cancellation in Ducts A Spanias and J Liu 21 pages ASU Report CRR 92044 Reported to ANVT February 1992 R 25 Fixed Point Implementation of the VSELP algorithm Phase 1 A Spanias M Deisher and P Loizou 14 pages ASU Report CRR 92001 Reported to Intel Corp June 1991 R 26 Transform Coding for Seismic Data Compression A S Spanias and S B Jonsson 104 pages ASU Report CRR 91001 Reported to Sandia National Labs for contract No 54 0899 July 1990 Patents Phase Compensation for Sinusoidal Transform Coding of Audio Signals S Ahmadi and A Spanias Arizona State University Submitted Invited Presentations e University of Cyprus Speech and Audio Coding Technologies Combined IEEE and Department of Computer Science Nicos
151. ollowing sub sections describe how to do so through a simple example 3 1 The Basics An example of html code that establishes and runs a simple J DSP simulation is listed below Dian nimi nian initia iin csi imc iiss inns inet nici ancients tiie mics inicio I inure anc ie ican icin rin cei icici sittin I mii inline niin ra iiinincniciintiniiaiiiniimiiniiininniiiiminitmint IIIA AEII iciainniuiinmia Diecast ein acini ccc nici ccc eas a eam amen recta A S crsseastuucussnesesesumessessauasssaravasaudeerevseeuastedravasasdearesseiaaedssusvasaudsavssseesaeissesvasserearsvceaaesseesvsuaeredsussunsasassesvuasisdsuvsenaaedssuuseuasdedeuscusdsaseusseuasdseruvssastesseussauasdsacuvsasnedsesssesasisecssacnlaratesuseesesid Note that the J DSP scripts are placed between the html applet tags namely lt APPLET gt which marks the beginning of the J DSP applet and lt APPLET gt that marks the end of this applet The applet tag lt APPLET gt also includes the following information it instructs the browser to load the applet by specifying its sub class name width 400 pixels and height 250 pixels In our case the user should always type the following lt APPLET CODE JDsp class WIDTH 400 HEIGHT 250 gt lt APPLET gt To create a J DSP flowgram with scripts you have to specify a set of parameters that establish the blocks and link them to form a flowgram These parameters are not associated with the DSP simulations but instea
152. onse of the LTI systems Moreover the students observe the filtering effects and get familiarized with the source filter configuration Six problems have been developed for this lab In problem 1 students are asked to simulate a digital filter using a given transfer function Figure shows an example simulation performed using J DSP In problem 2 the students are asked to design a digital oscillator In problem 3 a finite impulse response FIR filter is provided and the students are asked to observe the behavior of the system for different inputs In problem 4 the students study symmetric impulse responses In problem 5 the students compute the transfer function for various pole zero PZ representations In problem 6 they simulate cascade and parallel configurations FIGURE 1 J DSP BLOCKS USED IN SIMULATION OF DIGITAL FILTER IN LABI PROB 1 Lab2 Pole Zero Plots and Frequency Responses This lab deals with the effect of pole and zero locations on the magnitude frequency response First the relationship between the pole zero plot and the magnitude response of a system is covered Four problems are assigned in this lab In Problem 1 the students are asked to find the poles and zeros and observe the frequency response of a given filter In Problem 2 the students observe the variations in the frequency response by graphically moving the poles and zeros in the z domain or by manually entering their values In Problem 3 low pass and
153. onstant Ibl lt 1 M9 6 section M8 Speech blocks These blocks appear at the top of the simulation area Table of blocks Block notation Description Autocorr Computes the autocorrelation sequence of the input signal LPC Calculates the linear predictor coefficients LPC LPC Computes the LP coefficients LPC gt RC Converts the LP coefficients to reflection coefficients RC RC gt LPC Converts reflection coefficients to LP coefficients RC gt LAR Computes the log area ratio values LARs LPC gt LSP Converts LP coefficients to line spectral pairs LSP LSP gt LPC Computes LP coefficients from the LSP BW Exp Function to expand the bandwidth of the filter Inv TF Reciprocates the input transfer function Prep Fil Performs perceptual weighted filtering Autocom Lec cece teore R tree J Rca f Lec LsP f Lep Lec f Bw Ex f imete f Prep Fil M8 1 Block name Autocorrelation Notation Autocorr Please refer to section M7 block M7 1 M8 2 Block name LP coefficients Notation LPC Please refer to section M7 block M7 2 M8 3 Block name LP coefficients Notation LPC Please refer to section M7 block M7 3 Ms8 2 MS8 4 Block name LPC to RC Notation LPC gt RC Description This block converts the direct form LP coefficients a to reflection coefficients k The Levinson recursion algorithm is used to implement the LPC to RC conversion A check box option is provided t
154. oped 1 e general and concept specific In the general forms the students are asked to provide general qualitative and quantitative evaluation of J DSP and information collected includes comments on logistics accessibility convenience demographics academic standing etc In concept specific forms students evaluated each laboratory task with regard to its impact on learning specific DSP concepts Our newest assessment instruments assess learning attributed specifically to J DSP by testing DSP concepts before pre assessment and after post assessment the J DSP lab Assessment results have recently been submitted to the Online Evaluation Resource Library OERL From the evaluation document in Appendix C 1 we see that most of the students almost 90 or above agreed that the J DSP labs helped them understand the DSP related concepts In pre and post assessment Fig 2 students were given a quiz in class prior to performing the J DSP laboratory and the same quiz was administered after a lab was performed We observe that students have performed better in the assessment quiz after they have completed their J DSP labs Questions 1 and 2 assessed whether students can relate the pole and zero positions to the resulting magnitude frequency response A 13 1 improvement was realized in Question 1 and a 13 29 was realized in question 2 in the post lab assessment In Question 3 two polynomial functions are given and students are asked if their magnitude
155. or Award of the Year 2003 for his contributions in the J DSP project 6 3 Endorsements from Students e Y Song a former EEE 407 student writes the graphical user interface helps the user understand some basic DSP concepts better e C Panayiotou a former EEE 407 student writes the topic regarding the effect of windows on FFT spectra became clear with exercises and simulations conducted using J DSP e A Natarajan a former EEE 407 student I particularly liked that one could move the poles and zeros around the unit circle and observe the corresponding effect on the peaks and valleys present in the frequency response e Several anonymous student comments collected from J DSP web evaluations are itemized in Appendix C 2 7 J DSP Labware and NEEDS Courseware Criteria J DSP is not multimedia courseware per se and therefore some items from the NEEDS criteria may not apply in the J DSP NEEDS submission We classify J DSP as abware i e software that enables students to perform educational laboratories on the web We believe that this software must be evaluated by NEEDS because it represents a technology innovation that is intended solely for education and addresses the laboratory aspect of education that perhaps not many other submissions address Because this submission is not the traditional multimedia courseware we submitted all the materials supporting J DSP to demonstrate to the reviewers that J DSP is a comprehens
156. or on line lectures see reference 3 and J DSP manual in Appendix F This J DSP feature was developed by creating J DSP scripts for every function that are embeddable in html content The task was quite involved and comparable to developing a complete computer interpreter entirely completed for DSP 3 e Accompanying servlets to allow students to submit their J DSP lab reports for EEE 407 on the network 2 completed e Advanced DSP functionality for Speech Processing 9 Spectrograms 5 partially completed Other functions being developed in the NSF J DSP project include controls 4 image processing 8 and communications 10 Also as part of the NSF grant we are developing further the J DSP GUI In particular we are developing an interface to MATLAB see Appendix A Fig A 4 and we will develop a new GUI that allows seamless integration of J DSP functionality with streaming video audio Fig A 5 1 1 The J DSP Graphical User Interface J DSP has a rich suite of signal processing functions that facilitate interactive on line simulations of modern statistical signal and spectral analysis algorithms filter design tools QMF banks and state of the art vocoders All functions in J DSP appear as graphical blocks that are divided into groups according to their functionality Selecting and establishing individual blocks can be done using a drag and drop process Each block is linked to a signal processing function Figure 1 shows the
157. orrelation M7 5 M75 Block name Symmetric correlation Notation Sym Corr Description This block makes the autocorrelation lags r symmetric so that they can be used with the FFT block in order to calculate the power spectral density PSD Symmetry of the autocorrelation sequence around 0 is modified to symmetry around the edges Pin assignment Autocorrelation sequence r m Symmetric autocorrelation sequence r m Dialog window s x Symmetric Autocorrelation Name ia FFT Size Mj 256 Hoof Lage My 1101 Lags 1 A 50 100 150 200 250 Lag Values Close Update Help Close a Sym Corr dialog window and output values Script use Name symcorr Example code lt param name 3 value B3 symcorr 3 1 gt Equation s Implemented r N m r m where n FFT size and m number of lags For example if the FFT size N 8 and the number of lags is 3 then r 8 Pe 0 f 7 rall f 6 ra 2 and so on M7 6 M7 6 Block name Correlogram Notation Correlogram Description This block computes a PSD estimate by performing an FFT on the symmetric autocorrelation sequence Pin assignment Symmetric autocorrelation sequence rx m PSD estimate R k 2 ee ee Fs Dialog window s Correlograms Correlogram Name la FFT Size H 256 30 Magnitude dB 43 pi pi Close Updat
158. p you understand the J DSP environment A Yes 112 No 14 Did not view the demos at all 38 Q Establishing and connecting blocks is easy A Strongly Agree 74 Agree 73 Neutral 15 Disagree 1 Strongly Disagree O Q Do you like the idea of an Internet based simulation tool such as J DSP A Yes 151 No 13 Q Your occupation A High School student 1 Undergraduate student 60 Graduate Student 97 Faculty Instructor 1 Researcher 0 DSP practitioner 1 Technician 0 Engineering manager 1 DSP hobbyist 2 Other 1 Q Which continent are you logging in from A North America 151 South America 0 Europe 4 Asia 9 Australia 0 Africa 1 Q11 In your opinion is this type of on line lab concept beneficial for distance learning A Yes 157 No 6 Q The graphical interface of J DSP is intuitive and user friendly A Strongly Agree 44 Agree 89 Neutral 11 Disagree 1 Strongly Disagree 1 Q Would you consider using J DSP for small simulations apart from the lab exercises A Yes 128 No 17 Q9 Do you think with the help of a simple manual it is possible to learn the basics of DSP by using J DSP A Yes 127 No 19 Q10 Should the J DSP editor be established as a full fledged tool A Yes 132 No 14 B 2 Comments by students obtained from J DSP online assessment excellen
159. panil L l 1 0 0 Frequency Response Real Copyright 1996 2002 Arizona Board of FA For the latest on J DSP click the butta http vidsp asu edu 0 10 15 Linear dB Close Help Close Update Help Show Coef Java Applet Window Java Applet Window Java Applet Window 2 Export simulation in J DSP script 3 Copy and paste script into an HTML file 5 Deliver to students J DSP Future functionality Proposed to NSF Export simulations in MATLAB code TH J D gt Editor ew Help Demos SiMe ee CM Filter Blocks AE TETTA E Speech PZ Placement Pz Piot FIR Design ingbesign Kaiser Desion f tms f Sig Gen I peel SigGen L fi Coeff Junction Scri Export Filter Copy and paste this code MATLAB code generated by J DSP Edita Fre Resp a clear all close all posel Frequency Respon P ot Plot2 VARG filter NUM6 DENG VAR 4 Snd Player figure 6 freqz NUM6 DEN6 Quantizer figure 2 plot vAR2 Plot h 8 figure 8 plot VARB Filter f 6 NUM6 1 0 2 0 1 0 0 0 DEN6 PLO 2560 0 6 1 0 VARG filter NUM6 DEN6 VAR4 figurei freqz NUM6 DEN6 Java Applet Window t Window Java Apple Plot b 2 figqure 2 plot V aR2 Plot h 8 figqure s plot V AR6 1 Prepare demonstration in J DSP 2 Export simulation in MATLAB script 3 Copy and paste code into MATLAB
160. pass high pass stop band or pass band Wp Ws pass band and stop band edge cut off frequencies respectively Wp2 Ws2 second pass band and stop band edge cut off frequencies respectively for pass band filters PB SB pass band and stop band tolerances in dB Pin assignment Filter coefficients Dialog window s Parks McClellan Algorithm FIR Filter Parameters Mame ja Filter Type Low Pass COEFFICIENTS Cutoff Frequencies Filter Order 15 0 01297 igt 0 1 WS 0 2 0 01859 i 0 02974 r 0 04121 wpz 0 Wsz 0 0 05331 0 06412 0 07227 pa 20 se 200 a Tolerances dB Close Help Java Applet Window a Parks Mc dialog window Script use Name ParksMac Example code lt param name 3 value BO ParksMac 3 1 gt M6 7 M6 7 Least Mean Squares Aeon Notation LMS Block name Description Implements the sequential least mean squares adaptive filtering algorithm Pin assignment The signal to be modeled Reference signal Adaptive filter coefficients Dialog window s Sequential LMS Algorithm a x Sequential LMS Mame a Step 0 08 Order a LMS dialog window Script use Name LMS Example code lt param name 3 value B3 LMS 3 1 gt Equation s Implemented A new Set of adaptive filter coefficients is calculated for every new iteration in order to reduce the mean squared error The update equation
161. pdate for changes to take place What happens ans we have aliasing 1 e no signal Step 1 3 Create a sinusoid with frequency 1 37 amplitude 3 75 pulse width 40 What happens Count the number of samples in a period ans we have aliasing again signal makes no sense Step 2 Next we want to take a look at the Filter output in the time and frequency domain Set the values in SigGen as per step 1 1 Double click the Plot block and a new dialog window will appear You should again see the input signal because the filter is just letting the signal pass through unaffected since no coefficients have been set If you press the Graphs Values Stats button a table with the values of the signal appear In the first column you see the indices of the samples and the second column shows you the values Close the value dialog box Step 2 1 Let us now see the filter in action Keep the Plot window open to observe any changes Double click the Coeff block You should see the dialog window of Fig 6 Filter Settings Et x Filter Setting Mame b Numerator Coefficients B0 E10 Select display BO 1 0 B1 0 0 Be 0 0 B3 0 0 B4 0 0 B5 0 0 Be 0 0 Br OO Be 0 0 B9 0 0 B10 0 0 type by line Denominator Coefficients 40 410 AO 1 0 Al 0 0 A2 0 0 As 0 0 A4 0 0 Ao 0 0 AB 0 0 Ar 0 0 AS 0 0 Ag 0 0 A10 0 0 The filter coefficients are all zero 10 00 00 00 00 00 00 00 Oo oo oo Eest hol
162. plenty of support Forms are provided for assessment and feedback and student user responses and associated statistics are automatically posted using MS asp technologies on the web site These can be viewed on the J DSP web site http jdsp asu edu This software enables instructors to integrate simulations and Java animations in their web content The tedious Java programming of DSP tasks is replaced by object oriented visual programming That is the instructor can generate easily J DSP scripts that have been developed to activate J DSP simulations from HTML see Appendix A Fig A3 We view this as one of the greatest potentials for this tool Expansion of J DSP used to other areas such as communications controls and image processing has already started 11 We are also embedding this software in a new web course on DSP that we are developing 8 Software Design The software is reasonably engaging when used along with the lab exercises and does not contain stereotypes Although it would take sometime to load when connected with a POTS 33k modem once J DSP is loaded the execution is quite fast Loading is also very fast with ISDN DSL LAN and cable modems The software is continuously improved and we feel strongly that the DSP portion of J DSP is reasonably free of bugs One of the challenges in maintaining J DSP is dealing with the rapidly changing technologies of web browsers Therefore constant maintenance is essential 8 1 Learner
163. pplet Window no LL x Name fe Plot Magn scale linear dB Plot2 2 0D Snd Player 0 00 0 20 T Grid On Off Plot cont Axis Auto z Graph Values Stats Close Help Warning Applet Window Figure 7 Opening dialog windows 3 6 Help Dialogs The asterisk sign followed by any message generates a help dialog box containing the message An example is given below lt param name 15 value 1 This demo gives a rough idea of how to plot gt This line establishes the following dialog box in the J DSP editor JAYA SCRIPT DEMO WIKDOW DEMO HELP Warning Applet Window M10 11 3 An example The following example includes several of the J DSP scripts parameters The resulting J DSP editor window is shown in figure 8 lt applet CODE JDsp class width 400 height 250 gt lt param name numCommand value 14 gt lt START PARTS gt lt param name 0 value BO0 siggen 1 2 gt lt param name 1 value B1 filter 2 2 gt lt param name 2 value B2 plot 4 2 gt lt param name 3 value B3 coeff 2 5 gt lt END PARTS gt lt START CONNECTIONS gt lt param name 4 value C 0 4 1 0 gt lt param name 5 value C 1 4 2 0 gt lt param name 6 value C 3 3 1 2 gt lt END CONNECTIONS gt lt START OPEN DIALOGS gt lt param name 7 value O 0 gt lt param name 8 value O 2 gt lt param name 9
164. pt Electr Eng ASU Dec 2002 Algorithms for Beamforming S Miller Dept Electr Eng ASU Dec 2002 Analysis and Implementation of the MP 3 standard Rama Ramapryia Dept Electr Eng ASU Dec 2001 Analysis and Evaluation of G 723 1 Mugundan Narayanan Dept Electr Eng ASU Aug 2000 Mugundan is now with Intel Development of Java DSP Argyris Constantinou Dept Electr Eng ASU Dec 1999 Argyris is now with Globalsoft Development of Speech Processing Functions for JDSP Maya Tampi Dept Electr Eng ASU Dec 1999 Maya is now with Motorola Multichannel Noise Cancellation Anand Xavier Dept Electr Eng ASU Aug 1998 Anand is in Signalogic A Comparison of Vocoders for Cellular Systems with Emphasis on the Enhanced Variable Rate Codec EVRC Hiren Bhagatwala Dept Electr Eng ASU May 1998 Hiren is now with Qualcom Adaptive Modified Covariance Algorithms with Time Varying Gains Kyriacos Kitsios Dept Electr Eng ASU May 1995 Kiriakos is now with J amp P Analysis and Implementation of Speech Coding Algorithms Ted Painter Dept Electr Eng ASU May 1995 Ted is with Intel Endpoint Detection for Isolated Word Recognition Using Hidden Markov Models Greg Tucker Dept Electr Eng ASU May 1995 Greg is currently with Intel Normalized Frequency Domain Modified Covariance Algorithms Francis Tiong Dept Electr Eng ASU Dec 1994 Francis is currently w
165. quipment on the Project Development and Evaluation of Fixed Point Full and Half Rate GSM Coders PI A S Spanias Amount 36 000 00 Funds from ASU VPR EE TRC Date November 1991 e Research Incentive Award on the Project A Hybrid Model for Speech Coding PI A S Spanias Amount 16 000 00 Funds from VPR EE TRC Date August 1989 Labs and Facilities Developed e DSP Lab From Funds by Donations and Equipment Grants 200k 1989 2001 e Speech Systems Lab Anechoic Chamber From Funds by Donations and Equipment Grants 150k 1989 2001 e Distance Learning Lab From Funds by Donations and Equipment Grants 200k 1995 2001 Other Research Proposals e PI A S Spanias Speech Coding Libraries for Microcontrollers Amount 52 500 00 May 1995 May 1996 e PI A S Spanias with 5 CO PIs Development of Signal Processing and Communications Algorithms and Design of Small Antennas for a Campus Security System Sponsor Motorola GSTG Prime ARPA Amount 589 806 00 Jan 1994 Dec 1995 e CO PI A S Spanias with 20 others Intelligent Vehicle Highway Systems Research Centers of Excellence Program Sponsor Federal Highway Administration Amount 1 398 123 Aug 1993 Aug 994 e CO PI A S Spanias with 24 others Foundations of Intelligent Systems MOVE Army Research Office URI CRP 92013 Jan 1992 Dec 1995 1 662 143 00 e PI A S Spanias Speech Processing Based On Higher Order Statistics RIA National Science Foundatio
166. re is that instructors do not need to engage in tedious Java programming to embed simulations in their web content Other functionality was also developed to enable advance algorithms to execute in J DSP Furthermore assessments of leaming attributed specifically to the use of this software was carried on the web Functionality extensions formal assessment and implementation of advance algorithms are being 7 this year ina series of publications at JEEE ASEE FIE 03 I highly recommend J DSP educational eileen for the NEEDS award This software essentially enables DSP courseware with embedded on line simulations _ and establishes a new paradigm for running DSP simulations and laboratories over the internet Sincerely Fives C L Max Nikias 09 4 00 00 d OBP L c6P8 0F2 12 N30 AHL 40 391440 ONIYBRNIDNA OSN Wory wegg 80 002 01 UNf Apr 12 03 05 47a Yamacraw 443852140 a peog Georgia College of Tech Engineering EA School of Electrical and Computer Engineering june 10 2003 To whom it may concer _ This letter is written in support of the Java DSP J DSP educational software that has been nominated for _ the 2003 NEEDS award The J DSP on line simulation software was freely disseminated at the 2002 IEEE DSP workshop Georgia Tech obtained a free copy for testing and colleagues and students arc currently evaluating this software The idea with J DSP is to provide a free programming and visualization environ
167. requency using an FIR or an HR filter We observed an 8 7 improvement in the post lab assessment In Question 6 students are asked about the structure of poles given the coefficients For this question only 1 8 improvement was observed In summary we observed a 10 average improvement in lab 1 after J DSP was used Lab 2 assessment results Again we observe in all the questions students have performed better in the assessment after they have completed their J DSP labs Questions 1 and 2 assessed whether students can relate the pole and zero positions to the resulting magnitude frequency response A 13 1 improvement was realized in question 1 and a 13 29 was realized in question 2 in the post lab assessment In Question 3 two polynomial functions are given and students are asked if their magnitude responses are equal A 10 99 improvement is noticed here In Question 4 students are asked about the filter type i e low pass high pass etc of a given transfer function G Pre assessment Assessment of Lab2 m Post assessment O Improvement O gt 100 5 F 90 1 OF i 00 gt 80 8 L o oeg 70 2a g OF S Sig Oo do 2 404 Oo 5 O 5gs 304 a 2 59 2 20 10 0 Q1 Q2 Q3 Q4 Q5 Question Number FIGURE 4 B THE RESULTS OF LAB 2 ASSESSMENT We see that only 28 26 of students answered correctly in pre lab assessment and it is indeed poor but the percentage has increased by 44 71 in the post lab as
168. roject was an object oriented DSP simulation environment The J DSP simulation environment can run from any location and from any platform equipped with an internet browser The modem speed and computer requirements are modest and therefore it proved to be quite useful and cost effective for distance learning The original version of J DSP was developed in 1995 and results have been published at the 1997 JEEE International Conference on Acoustics Speech and Signal Processing ICASSP 97 Originally the software accommodated only basic signaling filtering and FFT based simulations The software received immediate attention from DSP educators and was posted on several web sites worldwide Since that time the software has been continuously augmented with new DSP functionality Several changes also have been implemented on the infrastructure of the program a The purpose of the J DSP software was non commercial and there has been no attempt to commercialize it In fact the software specializes on compact educational simulations and does not compete with products of The Mathworks or packages such as MathCad ASU started using this software in their undergraduate course and Andreas Spanias has developed and published on the web a suite of computer exercises whose aim was to provide hands on experiences that complement the theory taught in his DSP classes The program became very popular particularly with distance learners es Andreas Spanias and
169. rs can be entered They can be placed either graphically or manually Graphical manipulation of poles and zeros is achieved through buttons that allow placing moving and deleting Manually placing poles and zeros can be done either in square or polar form Pin assignment es ES Dialog window s PZ Placement Block Ei la Add poles zeraz Graphically Inaginaly 1 5 1 0 Feal 0 0 i NE 1 5 1 0 0 0 10 1 5 Close Update Help Show Coet Iwarning Applet window a PZ Placement dialog window Add Zera ESEE Script use Name pzplace Example code lt param name 3 value B3 pzplace 3 1 gt M6 2 M6 2 Block name __ Pole Zero plot Notation PZ Plot Description This block calculates and displays the poles and zeros of a transfer function in the z plane The block accepts filter coefficients at its input Pin assignment P Plot b a Filter coefficients gt Dialog window s Fole ero Plot ZEROS 0 5 0 59 0 48 0 59 0 18 0 19 Imaginary 0 18 0 19 0 19 0 19 0 2 0 19 0 51 0 62 0 51 0 62 POLES 0 39 04 0 39 0 44 0 0 0 0j 0 0 0 0j 0 0 0 0 0 0 0 0 0 28 043 0 29 0 43 OR Out of Range Warning Applet Window C a PZ Plot dialog window Script use Name pzplot Example code lt param name 3 value B3 pzplot 3 1 gt M6 3 M6 3 Block nam
170. s J DSP An Internet based Educational Tool for Digital Filter Experiments 1998 IEEE Symposium on Advances in Digital Filtering and Signal Processing Invited Victoria BC pp 57 61 June 5 6 1998 C 89 A Spanias and T Painter Network Applications of Speech and Audio Coding Algorithms Article in Multimedia Over the Broadband Network Business Opportunities and Technologies Invited International Engineering Consortium June 1996 C 90 A S Spanias A Pole Zero Adaptive Algorithm for Speech Processing JEEE International Phoenix Conference on Computers and Communications IPCCC 90 Invited Conf Proc Page 894 Phoenix March 1990 C 91 A S Spanias M Pattichis M Souropetsis D Petrondas A Schizas L Middleton Linear Prediction Analysis Applied in EMG C Pattichis IId International Conference on Quantitative EMG Invited page 35 Larnaca Cyprus June 1988 Non Refereed Editorials and Abstracts A Block Modified Covariance Algorithm for Spectral Analysis A S Spanias IEEE Transactions on Signal Processing Vol 48 2 p 2123 Aug 1992 A Spanias Signal Processing Society Conferences Vice President Column EEE Signal Processing Magazine editorial Sept 2001 Book Software Publishing submitted and under review e Perceptual Audio Coding Ted Painter and Andreas Spanias Prentice Hall To be published Dec 2003 e Speech Coding for Mobile and Multimedia Applications Andreas Spanias
171. s Hamme Sc Onder 70 Log Area Ratos LR 607T2 LARIT t0405 LARD 0 196 GARDHE DTA LARDA 0 4061 LARIS 02839 LARI 04002 LAR 0 2093 LARIE 0 0124 LARIS 00321 Close Update Help Jews Aopen Window a RC gt LAR dialog window Script use Name rc2lar Example code lt param name 3 value B3 re2lar 3 1 gt Equation s Implemented I k a aa eae i where k reflection coefficients LAR i Log area ratio 1 M8 5 MS8 7 Block name LPC to LSP Notation LPC gt LSP Description This block computes the line spectral pairs LSP from the LP coefficients Pin assignment LP coefficients a Line spectral pairs F 5 l D 5 A O OLS O amp 1 gt Dialog window s LSP Parameters Name 6d Order FIO 0 0555 F 0 073 F2 0 09684 FR 0 1619 F 0 173 x mo FS 0 2468 FIB 0 3135 FP 0 3317 FB 0 3905 FS 0 4325 Imagina LSP 5 guay 0 0 1 1 5 Sh 0 Close Update Help a LPC gt LSP dialog window Java Applet Window Script use Name Ipc2Isp Example code lt param name 3 value B3 Ipc2Isp 3 1 gt Equation s Implemented A z z A z ESA A z z A z l z Each polynomial has five conjugate roots on the unit circle and they alternate each other The sum polynomial F z is given by F z The difference polynomial F z is given by F z M8 6 M89 Block name LSP to LPC Notation LSP gt L
172. se of appropriate functions Parameters in the blocks can be edited through dialog windows Blocks can easily be manipulated edit move delete and connect using the mouse Execution is dynamic and therefore any change at any point of a simulated system will automatically take effect in all related blocks All dialog windows can be left open to enable viewing results at more than one point in the editor TE J DSP Editor File View Help Demos Swi AO Mm Basic Blocks PLANNED FUNCTIONS Speech DISCLAIMER Sig Gen For Fen Java Applet Window Fig 1 J DSP simulation environment HEQJ DSP Editor E view Help Demos DENI apem Save as Intro to JOSEP DSP Editor die Demos FFT Dermo COEFF Gemo Pole erd Demoa Export as Script import from Script Disclaimer Close e I i j B yi Grid Onlot p E Spectral Est Demo LFE LSF Demo Eq J DSP Edita File View Help Demos EXISTING FUNCTIONS fBasic Blocks PLANNED FUNCTIONS Speech I block set selection Hasic Blocks Ha sic HI oc ks Arithmetic Speech Speech Speech Ill Analog comm Dig Comm O Basic Blocks EA 2D Filters Junction Press buttons to establish blocks 2D Transforms Statistical DSP Filter Siz Ger FFT Flot 4 b E ond Player Cluantizer Fig 2 Signal processing functions and menus in J DSP M1 2 1 2 Working with J DSP The easiest way
173. sessment This is significant In Question 5 students are given four choices regarding the change of magnitude response with displacement of the zeros with respect to the unit circle A noticeable improvement of 25 62 in post lab is evident Lab 3 assessment results On the average around 11 improvement is evident in each question in the results of pre post assessment of the lab 3 assignment regarding the design of FIR and HR filters In Question 1 the 10 frequency response of an ideal low pass filter is given and the students are asked to specify whether the filter is FIR or IR and whether it is causal or non causal A 11 9 improvement of the student s performance is evident Questions 2 and 3 are related to the characteristics of linear phase FIR filters and 15 46 and 22 37 improvements are observed respectively These improvements reveal that the lab clarified further the student understanding of linear phase filters In Question 4 the students are asked to choose an optimal filter design method from four choices namely frequency sampling Parks McClellan Kaiser window and Fourier series method A 12 08 improvement is observed Question 5 addressed the reasons for which one may choose an FIR filter Question 6 is set to assess the student s awareness on the constraints posed on the magnitude frequency response by certain linear phase designs In Question 7 four impulse responses are given to the students and are asked to choose t
174. t simulation of DSP practical work with high level of simplicity and flexibility bravo good product More functions such like save load cut paste Very Good Program I think it would be nice if we could save a block diagram along with coefficients etc and get back to it later and work Nice concept This is very easy to learn for beginners A great concept that should be used in all areas of study It is a great concept and it has been programmed very well This is a good program to study DSP It s totally good Overall I think it s quite an intuitive and interesting tool which gives the student a real feel of the subject This is good but will be better when more capabilities are built in such as being able to save diagrams and hear sound better and longer This is a good tool However a better help menu will be welcomed Participating University and Industry Test Sites As part of our dissemination and testing activities we have already contacted and obtained preliminary agreements from faculty in several undergraduate programs as well as engineers in relevant fields in the industry Universities that have already agreed to evaluate the modules and all relevant materials include Georgia Tech University of Southern California University of Maryland University of Minnesota University of New Mexico University of Texas Austin University of Texas Dallas University of Central Florida Northeastern Uni
175. tal signal processing and speech coding He has developed a software concept called Java DSP that is intended for use in undergraduate and graduate education http jdsp asu edu Andreas Spanias has been the principal investigator on research contracts from Intel Corporation Sandia National Labs Motorola Inc and Active Noise and Vibration Technologies He has also consulted with Inter Tel Communications Intel Corporation Motorola Texas Instruments and the Cyprus Institute of Neurology and Genetics He is member of the DSP Committee of the IEEE Circuits and Systems society and has served as a member in the technical committee on Statistical Signal and Array Processing of the IEEE Signal Processing society He has also served as Associate Editor of the IEEE Transactions on Signal Processing and as General Co chair of the 1999 International Conference on Acoustics Speech and Signal Processing ICASSP 99 in Phoenix He served as the IEEE Signal Processing Vice President for Conferences and the Chair of the Conference Board He served as a member of the IEEE Signal Processing Executive Committee and as Associate Editor of the IEEE Signal Processing Letters He is currently serving as a member of the IEEE SPS Publications Board and as a member at large of the IEEE SPS Conference Board He has been Chair of the IEEE Communications and Signal Processing Chapter in Phoenix and is a member of Eta Kappa Nu and Sigma Xi Andreas Spanias is recipient of the 2
176. than 90 students are comfortable with the DSP related topics after completing the J DSP labs In the evaluation they were also asked if they received additional knowledge by performing the lab assignments in addition to the lectures A few comments are presented to give a picture of student s impression In Lab 1 some of them stated that the lab helped understand better the concept of cascaded and parallel configurations 14 Should the LOSP editor be established as a full fledged tool Would you consider using J DSP for small simulations apart from the lab exercises In your opinion is this type of on line lab concept beneficial for distance learning Do you think with the help of a simple manual it is possible to learn the basics of DSP by using J DSP More than an hour ot 15 minutes or less 36 How long did it take to get familiar An hour with the basics of the 24 J DSF environment Half an hour 32h FIGURE 6 USER FEEDBACK REGARDIND J DSP EDITOR USED IN DSP LAB Other comments included confirmations from students that J DSP helped them understand better the issues related to filter design and the fact that pole and zero locations relate to the frequency response TABLE I STATISTICS BASED ON USER EVALUATION OF LABS 1 4 Evaluation questions Lab No Strongly Agree Agree Neutral Disagree Strongly Disagree 15 Improvement of your understanding of the con
177. the same as double clicking on a block in the J DSP editor frame The number of the block has to be specified here This is the number that we assigned to the block when it was established For example O 1 opens the dialog box of the second block of the flowgram because 1 is the number of the second block The complete line is lt param name 9 value O 1 gt Now we are ready to write a simple program to connect two blocks in J DSP and open their dialog boxes to observe input and output plots Here is the necessary code The above program establishes a Sig Gen block in the 3 1 position and a Plot block in the 5 1 position of the editor frame The lt PARAM gt tag in line 4 connects the two blocks by connecting Sig Gen block s 4 pin and Plot block s 0 pin The last 2 lt PARAM gt tags in lines 5 and 6 open M10 10 the dialog boxes of the SigGen block block number 0 and the Plot block block number 1 respectively Observe that the lt PARAM gt tag with numCommand as PARAM name has VALUE 5 because there are 5 general lt PARAM gt tags in this program Opening the HTML file and starting J DSP will give the following flowgram as show in figure 7 Signal Generator ce x Signal Generator Name fa Signal Preview Signal Rectangular 2 Gain fi 0 Pulsewidth 20 Sig Gen Periodic Period fio a a aa 10 20 30 Time Shift fo Samples Close Update Help Warning A
178. tion Aana i 0 0 0 0 1 0 simulating Vocoders CELP LPC Fundamental DSP ene Functions FFT IFFT MIDI tone generator x ius Windowing etc lt gt HMM Simulation IDR Training HMM Help 1 HMM Training for DIGIT 2 0 04 0 0 0 0 Images Amplitude x scale linear dB f Tonefone frame Toners fames 0 frequency samples 128 Grid SameXaxis Plot cont Axis x Close Update Help Reset Close Help Java Applet Window t Tonefall frames Recording Seamlessly embed J DSP simulations in web content TE j D gt Editor Sig Gen SigGen L Coeff Junction Pict sapplet CODE JDsp class width 400 lt param name numCommana value Plot2 lt l START PARTS gt lt param name 0 yalue B0 siggen 0 1 lt param name 1 yalue B1 filter 1 1 Sng Player lt param name 2 yalue B2 plot 2 1 gt lt param name 3 yalue B3 freqresp 1 lt param name 4 yalue B4 pzplace 1 sl END PARTS gt Quantizer lt START CONNECTIONS gt lt param name 5 value C 0 4 1 0 lt param name 6 yalue C 1 4 2 0 gt lt param name 7 yalue C 1 3 3 2 gt lt param name 8 yalue C 4 3 1 2 gt sl END CONNECTIONS gt lt L START OPEN DIALOGS gt lt param name 9 value 0 3 gt lt param name 10 value 0 4 gt lt l END OPEN DIALOGS gt
179. tor Unit Action Potentials Preliminary Findings Proc IEEE Intern Conf of Engineering in Medicine and Biology pp 1465 1466 Paris France Oct 1992 Q Shen and A S Spanias Frequency Domain Adaptive Algorithms for Active Sound Control International Noise and Vibration Control Conference pp 207 212 St Petersburg Russia May 31 June 3 1993 Q Shen and A S Spanias An Optimal Block Adaptive Algorithm For Active Control of Sound Noise Control 1993 NOISE CON 93 pp 231 236 Williamsburg Virginia May 2 5 1993 Q Shen and A S Spanias Frequency Domain Adaptive Algorithms for Multi Channel Active Sound Control Second Conf on Recent Advances in Active Control of Sound and Vibration pp 755 766 Blacksburg April 28 30 1993 Q Shen and A S Spanias A Multichannel Block Adaptive Algorithm For Active Noise Control 1992 International Congress on Noise Control Engineering Inter Noise 92 pp 353 356 Toronto July 20 22 1992 A S Spanias G Lim P Loizou and M Deisher Block Modified Covariance Algorithm Proc International Conference on Acoustics Speech and Signal Processing ICASSP 92 Vol 5 pp 529 532 San Francisco March 1992 Q Shen and A S Spanias Time and Frequency Domain X Block LMS Algorithms for Single Channel Active Noise Control Second International Congress on Recent Developments in Air and Structure Borne Sound and Vibration Congr Proc pp 353 360 Auburn Marc
180. trial Training Authority of Cyprus in association with the Cyprus Institute of Neurology and Genetics This one day course was given twice at the Makarios Hospital Nicosia Cyprus on August 5 1991 and June 17 1992 Support material 200 pages Speech Coding for Mobile and Multimedia Applications This course 16 hours was sponsored by the Professional Development Center of Arizona State University ASU and was held March 17 18 1994 An expanded version 24 hours of this course was also given May 18 20 1994 to Intel engineers in Chandler Also held June 1995 and June 1996 1997 1998 1999 2000 Shorter version given as tutorial at IEEE ISCAS 95 Support material 300 pages MATLAB for DSP Applications This short course was given in 1997 1998 1999 and 2000 Support Material 300 pages ASU Committee Service Department Committees EE Graduate Committee Chair 1996 97 2000 present Systems Area Committee Dept EE Member 1988 present Chair Spring 1995 Chair 1998 present Department Personnel Promotion and tenure Committee member 1993 96 1999 2000 Department Executive Committee 1993 95 Undergraduate Committee Dept EE member 1988 89 1989 90 1990 91 1991 92 Chaired Spring 1990 the Systems Sub Committee Dr Spanias Dr Crouch Dr Grondin of the Undergraduate Committee responsible for the review of ECE301 EEE302 EEE303 EEE405 EEE406 EEE407 EEE480 and EEE482 Several Faculty Search Committees Member and Chair
181. u The project involved development of software for the Vector Sum Excited Linear Prediction Algorithm Sponsor Intel the student was paid from one of my Intel contracts Internet based Education This project started in 1996 and is being funded in part by the CIEE FGIA program 28
182. use Name dsample Example code lt param name 3 value B3 dsample 3 1 gt Equation s Implemented y n x nM x n input signal y n output signal M down sampling factor M3 7 M3 7 Block name Up sampling Notation U Sampling Description Up samples the input signal by an integer factor L L is allowed to take values from 1 to 10 Pin assignment Input signal Up sampled signal O a a oe Dialog window s Up Sampling Up Sampling Rate 1 10 2 warning Applet Window a U Sampling dialog window Script use Name usample Example code lt param name 3 value B3 usample 3 1 gt Equation s Implemented y n x n L x n input signal y n output signal L up sampling factor M3 8 M3 8 Block name Convolution Notation Convolution Description This block performs a convolution operation between its input signals Pin assignment Input signal x n Input signal x2 n Convolved signal y n cs Dialog window s None Script use Name conv Example code lt param name 3 value B3 conv 3 1 gt Equation s Implemented N 1 yn x n x 0 y n Dx m x n m m 0 x1 n input signal x2 n input signal y n convolved signal M3 9 Section M2 General blocks These blocks appear at the left of the simulation area Table of blocks Block notation Sig Gen Sig Gen L Co
183. ut signal y n St Dialog window s None Script use Name multiply Example code lt param name 3 value B3 multiply 3 1 gt Equation s Implemented y n x n i x n x n input signalat pin 1 X2 n input signal at pin 2 y n output signal M4 2 M4 2 Block name Logarithm base 10 Notation Logl0 Description This block calculates the common base 10 logarithm of the input signal Pin assignment Input signalx n Output signal y n 2 E 4 5 Dialog window s None Script use Name log10 Example code lt param name 3 value B3 log10 3 1 gt Equation s Implemented y n log o x x n input signal y n output signal M4 3 M4 3 Block name Natural logarithm base e Notation In Description This block calculates the natural base e logarithm of the input signal Pin assignment Input signalx n 2 Output signal y n PS 2 ek re ee ON d Dialog window s None Script use Name In Example code lt param name 3 value B3 In 3 1 gt Equation s Implemented y n log x x n input signal y n output signal M4 4 M44 Block name Exponential Notation exp Description This block calculates the exponential of the input signal Pin assignment Input signalx n Output signal y n ek re 3 ee ON d Dialog window s None Script use Name
184. utral Disagree MO Strongly Disagree Q11 Can you suggest an exercise along the lines of this one gi Q12 Please suggest possible improvements relative to this lab such as redesigning of dialog box of a block what to add or delete etc General evaluation questions This is a list of the general evaluation questions students were required to complete after using J DSP This list is also available online at http jdsp asu edu for all other users Part 1 of 3 Q1 What type of Internet access did you use ES 28 8 56 6 modem L DSL cable modem Li LAN Q2 How long did it take to get familiar with the basics of the J DSP environment La 15 minutes or less La Half an hour L An hour L More than an hour Q3 How would you rate the JDSP concept on a scale of 1 bad to 10 excellent E Q4 Did the demos help you understand the J DSP environment s Yes s No La Did not view the demos at all Q5 Establishing and connecting blocks is easy L Strongly Agree La Agree La Neutral La Disagree L Strongly Disagree Q6 Do you like the idea of an Internet based simulation tool such as J DSP La Yes Li No Q7 Overall do you think its worthwhile using J DSP as opposed to MATLAB for small tasks s Yes s No Q8 Your occupation r High School student Undergraduate student Graduate Student Faculty Instructor Researcher DSP practitioner Technician Engineering manager DSP hobbyist Other C1 G2
185. value O 3 gt lt END OPEN DIALOGS gt lt l START PART PARAMATERS DO NOT MODIFY gt lt param name 10 value P0 30 10 0 1 0 0 9 0 0 0 2 a Triangular No null gt lt param name 11 value P1 gt lt param name 12 value P2 c linear Magn cont false gt lt param name 13 value P3 1 0 1 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 2 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 d gt lt END PART PARAMATERS gt lt applet gt M10 12 4 J DSP Editor File View Help Demos _SNR Statistics Window Mier D Sampling U Sampling Convolution Sig Gen SigGer L Coeff Junction Filter cale i lnea dB Snd Player Signal Generator Mame fd Sign Numerator Coettn Select H a y j ame coou i display Signal Triangular F Be 0 0 By tpe GridOnO Plot cont Axis Auto by line Denominator Coe Graph Values S tats Close Help Gain E ade a Warning Applet Window AB O00 AP Pulsewvidth 30 Masel Se inset OBI ote 00 00 00 00 00 00 00 oo 0o ASE 3 f C ADAID Time Shift a Close Close Update Help mee a a a Boe SE Figure 8 The result of the example script M10 13 Section M9 Audio effects blocks These blocks appear at the top of the simulation area Table of blocks Block notation Description DTMF tones Generates Dual Tone Multi Frequency tones MIDI Generates MI
186. versity We also have agreements with several other universities that have previously obtained the J DSP software and agreed to evaluate future updates Universities that agreed to establish J DSP web sites and test lab content include University of Kent Marquette University Stevens Institute of Technology Georgia Institute of Technology Blekinge Institute of Technology Drexel University University of Nebraska Cal Poly Pomona University of Detroit Mercy University of Pennsylvania Iowa State University University of New Brunswick University of Alabama Huntsville Industry test sites include Texas Instruments Motorola Intel Semy Engineering Rice University Massachusetts Institute of Technology University of Akron University of Connecticut University of Puerto Rico Clemson University North Carolina State University Ecole Nationale Polytechnique Algeria Bogazici University Turkey University of Rhode Island General Dynamics Honeywell Nokia We also add that we have a commitment from a DARPA program manager to disseminate our materials to relevant federal entities J DSP dissemination efforts This is a list of people who obtained the J DSP software for evaluation and or use at their own academic institutions NAME NABIL ZAKRIA MICHAEL JOHNSON A ELLCONY HONG MAN DAVID ANDERSON LARS OLOF LARSON JAN MARK DE HAAN RZHANG E N BIDEN ALAN FELZER SHUVRA DAS MITCHELL LITT JOHN PROAKIS JULIE DIC
187. viewing and interpreting spectra of several bench mark signals m Pre assessment Assessment of Lab4 m Post assessment o Improvement y N 97 5 O ol ae 0 i Pins LO 80 N N N 70 oe 60 os 32 k Ta o 62 9 2 0 SESE gy 9326 S 30 30 a A 90 4 10 0 Q1 Q2 Q3 Question Number FIGURE 4 D THE RESULTS OF LAB 4 ASSESSMENT Lab 5 assessment results I2 Lab 5 is on multi rate signal processing and QMEF banks Figure 5 shows the assessment results of Lab 5 In Question 1 the students are asked to describe how spectral domain signatures are affected by down sampling and up sampling m Pre assessment Assessment of Lab5 m Post assessment o Improvement s0 Qe QR 70 52 60 Z LO Pa 8 L X 9 ct T 0 TE eo T oF 30 cow N A D N o fa 20 rao 2D Tm 10 4 z gt m CO 0 i D u Q1 at stion Nun er Q4 FIGURE 5 THE RESULTS OF LAB5 ASSESSMENT Question 2 is about the use of a reconstruction filter for interpolation Question 3 is related with the placement of a quantizer in a QMF bank An average improvement of 8 is observed in this lab This laboratory was also one that the students benefited from in that they gained hands on experience on filter banks which are now common in MP3 players and MPEG video compressor S 6 GENERAL ASSESSMENT In this assessment students gave their general subjective opinions on J DSP and provided us
188. xplained in the following sub sections 3 2 Creating parts One important instruction that can be given to the J DSP editor is to create a new part or block as parts are often referred to Therefore B represents a new block and is followed by a number which specifies that this is the block of the flowgram This number is also used to refer to the part when performing other tasks with scripts like opening its dialog box The block name follows with the coordinates of the editor frame in parentheses For example BO siggen 3 1 means that the first block of the flowgram is a SigGen block and is to be placed in the coordinates 3 1 The editor frame grid is shown below Each box is approximately 50x70 pixels 012 gt 0 Ce T HI Editor frame grid Hf TIL T tte tte TT EET ET TE TT i Here is the lt PARAM gt tag for this case lt param name 0 value BO0 siggen 3 1 gt Now we are ready to write a short and simple program to establish a block in the editor frame ececsesesecersenecenss enreeeeeeeeeeeeeeeeeeeeseeeeeeeeeeeeeeeeeeeeeeeeeaeseeeeeeeeeeeeeeeeeaeseeeeaeeeeeeeeeeeeeeeeeeseeeeeeeeeeeeeeeeeeeeeeeeeeeeaeseeeeaeseeeeaeseeeeeeseeeeeeeeeeeuseeeeeuseeeeeeeeeeeeuseeeeseeeeeeeeseeeeeeseneeeeseeeeeseeeeeeseeeeeuseeeesueeeeeseeeeeeeeseeeeeeseeeeeeseeeeeseeeeeeseeeeeeeeeeeeneiy Observe that the first lt PARAM gt tag in line 1 with numCommand as PARAM NAME has VALUE 1 because there is only one general lt P
189. y generate J DSP scripts A user simply needs to create the desired flowgram using the familiar drag and drop procedure of the editor Then by selecting File and Export as Script the user obtains the script ready to use in an HTML file This automatically generated script also includes all the blocks parameters exactly as they where defined when the simulation s flowgram was saved providing full control over a saved simulation Section 2 1 contains step by step instructions on how to create your own J DSP scripts 2 1 Scripts in a flash A few steps is all that is needed to get a J DSP script and save it over the Internet for others to use Step1 Start the J DSP Editor and create a flowgram as desired Don t forget to set the desired parameters to all the parts Step 2 From the main menu select File and then Export as Script as shown in figure 1 A new window appears that contains the code describing the simulation as shown in figure 2 This will be referred to as the J DSP script window Note that by default the window presents only the applet code which can be placed inside the body of an existing HTML file If the entire HTML file is needed select Applet in HTML code from the drop down box of the script window 4 J D5P Editor View Help Demos ne ac Block COBEN s Save SAVE FS Export as Script Import From Script Close uit Loet Figure 1 Selectin

Submission Docs

Contents

Download Pdf Manuals

Related Search

Related Contents