Home

Clinical Applications Manual

1. _ cf of lt clo rrav rrav save penr Paste ane sEL 0 475 SOUNDSCOPE IS A PRODUCT OF G W INSTRUMENTS INC SOMERVILLE MA 16 3 mSec Div 2712 2921 3130 3338 3547 3755 m x 299 msDiv x 75 9 mio CLIENT STIMULUS PLAY RESP HIT cushion eyes on 8 morph stabilize 8 37 5 62 5 75 0 100 0 500 1000 1500 2000 2500 3000 3500 4000 msec 500 msec Div Fi INTRODUCTION SoundScope was designed to be used with a Macintosh computer It works best on models that are at least as powerful as a Macintosh IIfx equipped with a 1 4 meg floppy drive and at least 8 megs of RAM The monitor you use should be set to color and 256 settings under Monitor in the Control Panel It is assumed that users will have some familiarity with the Macintosh i e understand its desktop know how to find and open up programs and understand the func tions of a mouse If you are a true novice you might want to try something like Professor Mac a software tutorial from Individual Software Inc in Pleasanton CA Basic operations of the Mac are reviewed extensively in this excellent primer After you put the SoundScope Clinical Manual demo disk into your computer s floppy drive open the Read Me file and follow its instructions to uncompress the files you ll need to carry out the exercises described in t
2. _ _ANaLyze muro LOAD PLRY tens enim RUG 335 MIN 129 anes Sanna SPEECH SEL SEGMENT SPEECH SVLLABIC RATE 7 DURATION SEL 0 01065 SEG 1 238 pun 5 748 ses EDIT mam STD DEV 143 58 SvL 000 7500 msec 500 msec Di Waveform compression and expansion FO plot is displayed in this window Range of frequencies shown can be adjusted with this button When used as RANGE CALCULATOR this button is ON and DIADO RATE is OFF FO controls are on right of control panel expansion permits samples per one cycle to be seen Expansion from 500 msec Div display above 9 9 SSS RATE RANGE CALC 6 14 S 20 msec Div reveals FO plot cape e ate Sore teer qeu as series of red dots Se ANALYZE AUTO LOAD PLAY SPEECH SEL REcogD pray sav SET REC eatace fevera Sto now m amp 2 Gwi EH Expansion also reveals green Eme Mn rrr seem mam Cal 2 255 KH Lez se 5 210 215 ANALYSIS SAY IU JGNL J PARAMETERS une curs ve SEGMENT SPEECH 0 01065 SEG 1 238 our 5 748 ees STD DEV 143 58 vertical lines which correspond to peaks in the FO plot SoundScope uses these to la determi
3. Appendix C page 29 Between each spectrogram win dow and its associated waveform window is yet another very short text window By pointing at it with the cursor and clicking the mouse you can type in information regarding the material displayed 17 At the top of the display is a control panel with a variety of buttons and indicators which are the primary tools for Reconstructor Each spectrograph window has 5 buttons along the right margin which are used to control functions related to spectrographic analysis Finally there is a button labeled H in the upper right corner of each wave form window When this button is clicked the time scale is auto matically adjusted to display the entire waveform exactly within the bounds of the screen In order to try out Reconstructor you will first need the utterance saw pencil house from the demo stimuli provided Click on LOAD A Reconst Function 3 When a dialog box appears listing items which can be opened find Demo Stimuli 2 and double click on it When it opens find saw pencil house and double click on it We recommend that you load the utterance on both the upper A and lower B screens So click on LOAD B Reconst Function 4 and perform the same operation to load the utterance in the lower display Click on H in the upper right corner of each waveform window to assure that you re seeing the entire utterance Check to see that the time scale for each spectro
4. If you place the vertical line of this icon over one of the segment markers and use the same press drag release tech nique you used earlier to select speech you can reposition the seg ment marker to the left or right of its current position Remember that a first click in a deactivated window is necessary to activate it Try isolating pencil or house using this method by moving the two segment markers to points at the beginning and end of the word When you ve isolated a portion of the utterance in this way click PLAY SEG A Reconst Function 9 in the control panel If you click in the box where INTERVAL SELECTED is now show ing you will change it to INTERVAL SEGMENT A and all the functions indicated by the buttons beneath it i e CUT AMP NOR MALIZE COPY DUR and PASTE Reconst Functions 15 420 will now apply to the segment of waveform A you ve just created Notice that clicking to INTERVAL SEGMENT A caused the duration for the interval to appear in the DURATION indicator An alternative method of moving the segment markers to bound a portion of an utterance is to first select the portion as described earlier and then click on A SL gt SG Reconst Function 11 if the selection is in waveform A or click on B SL gt SG Reconst Function 12 if the selection is in waveform B Clicking on one of these buttons will always cause the segment markers to move to the begin ning and end points of the selection This
5. SGSMP included in the Demo Stimuli 2 file it should be possible to construct stimuli similar to the ones described here using the tools under SELECTION Saute the sample stimulus contains components for cushion of air as well as for morph and stabilize You could also create something like eyes on from t Extract a portion of t COPY Stim Gen Func tion 16 it then PASTE Stim Gen Function 17 it onto an empty area on the waveform and then paste a second copy next to the first Now you can experiment with the interval between them with CUT Stim Gen Function 15 until you re satisfied with the result You can also vary the amplitude of the new stimulus with AMP Stim Gen Function 19 which allows you to increase or decrease the intensity of a selection The spectrogram display at the bottom of Stimulus Generator is useful for visualizing the stimuli you re constructing in more detail You can link waveform A B C or D to the spectrogram display by pressing on the square menu button just under Spe to the right of the spectrogram window and selecting the stimulus A B C or D you want to examine spectrographically Adjust the time scale at the bottom of the spectrogram to approximate that of the window that contains the waveform you ve chosen to analyze Click on Cal to draw the spectrogram The CALC DUR Stim Gen Functions 20 and 21 control under SELECTION can be used to make time measurements if
6. a pre recorded speech utterance into the wave UTT The pre recorded utterance must first be stored on the computer s hard disk using the SAV UTT function See Common Function 10 SAV UTT DURATION This control loads the values of the current durations of the Selected wave if any and of the wave segment bounded by the two vertical segment markers into the displays SEL and SEG respectively The two displays are located to its immediate right See Common Functions 13 SEL and 19 SEG NAN Not a Number This control is only operable when one of the two EDIT functions See Diado Function 8 and Pitch Function 9 is in the DRAW mode In Diado DRAW click on NAN and OK and then click on F1 F2 or F3 in the left margin of the spectrogram Select the portion of the corresponding formant trace you want to discard and click again on NAN In Pitch Analysis click on NAN select the portion of the pitch plot you want to eliminate and click again on NAN Unlike the value zero a point with the value NAN is ignored during data analysis operations Consequently if incorrect formant or pitch values are changed to NAN values subsequent analyses will ignore the unwanted data points See Pitch Function 3 ANALYSIS PLAY This control allows the user to play the speech in the wave UTT PLAY SEG Play Segment This control allows the user to play the speech in the portion of the wave UTT that is bounded by the two vertical markers PLAY SEL Pl
7. calculated as well as the statistical measures shown in the accompanying displays See Pitch Function s 4 DUR 5 AVG 6 STD DEV 7 MIN and 8 MAX When the EDIT control is set to DRAW and ANALYZE is clicked the statistics are recalculated based on the current pitch track THE PITCH TRACK IS NOT RECALCULATED The ANALYZE function will not operate if the EDIT control is set to LOG By clicking on SAV TO JNL See Common Function 14 SAV TO the pitch statistics along with the date time and the client s name can be recorded to the Journal for later archiving to disk 4 DUR Duration of FO Analysis This Display shows the total duration of the speech for which pitch values were calculated Thus if the segment duration See Common Function 19 SEG is 1 5 sec but this display reads 0 5 sec this means that there was a total of 1 sec for which pitch values were not calculated 5 AVG Average of F0 This display shows the average of the calculated pitch values 6 STD DEV Standard Deviation of F0 This display shows the standard deviation of the calculated pitch values 7 MIN Minimum of This display shows the smallest of the calculated pitch values 8 MAX Maximum of F0 This display shows the largest of the calculated pitch values 9 EDIT This control allows the user to change the cursor from its normal function EDIT to a DRAW or a LOG function This setting is a three way toggle that alternates between EDIT DRAW a
8. into the utterance of waveform A starting at the first segment marker for waveform A 0 00 OO BONS 29 Any questions comments etc about the instruments or manual can be directed to Rebecca Leonard Dept Otolaryngology Head and Neck Surgery Univ California Davis Medical School 2500 Stockton Blvd Sacramento CA 95817 We would appreciate your feedback The following demo packages may be ordered from GW Instruments Inc 35 Medford Street Somerville MA 02143 Tel 617 625 4096 FAX 617 625 1322 SoundScope Clinical Demo 25 SoundScope Demo 10 Clinical Manual spiral bound SoundScope demo software amp documentation Clinical instruments amp sample waveforms on disk Includes SoundScope User s Manual Rate Range Calculator Stimulus Generator sample waveforms and numerous instruments Speech Reconstructor 1 Channel Analyzer 2 Channel Analyzer SoundScope demo software amp documentation Segment Analyzer 2 Channel Editor etc For Macintosh order GWI SoS cm demo For Macintosh order GWI SoS demo For Power Macintosh order GWI SoS cm demo PPC For Power Macintosh order GWI SoS demo PPC
9. of interest expand the range of frequencies displayed to visualize higher frequency components Once you ve characterized the spectrographic details click on ANA LYZE DiadoFunction 3 to calculate the speaker s repetition rate for ta If you use SEGMENT SPEECH Common Function 18 be sure to check the segmentation To get an accurate rate you may have to move the markers manually and set the MANUAL AUTO MATIC control Diado Function 6 to MANUAL Set UTT TYPE count the syllables within your new segment boundaries and re ANALYZE Diado Function 3 How does the rate for ta compare to pa If one or the other is faster it may reflect greater integrity of one articulatory structure i e lips versus tongue over the other Substantial differences in rates may also provide insights regarding the speaker s ability to alternate between structures i e lips and tongue jaw to produce the conso nant and vowel in pa as opposed to using the same primary struc ture i e tongue jaw for both consonant and vowel in ta and ka This kind of information may be helpful to the clinician in planning strategies and ordering therapy objectives for particular speakers Now open KA KA KA DYS and scrutinize it just as you have pa and ta Be especially careful in the analysis of diadochokinetic rate for ka There are a number of additions in the utterance which reflect phonatory and respiratory irregularities that you won t wan
10. off and on or generate a range of frequencies and intensities may be useful The dual purposes of Calculator are reflected in the control panel located in the middle of the display see page 4 Those functions related to syllable repetition tasks are positioned on the left while those having more to do with voice analyses are on the right All controls common to both instruments for loading storing and other wise manipulating utterances have been positioned in the middle of the control panel Controls which are in bold print will when acti vated cause the named operation to be performed words not in bold generally serve as labels or indicators To try out Calculator first depress the LOAD SPEECH button and open the DEMO STIMULI 1 folder If you own SoundScope of course you can also RECORD your own utterances after setting record levels with SET REC LEV ELS Double click on PA PA PA DYS You will see the wave form of the utterance loaded into the window at the bottom of the screen Each time you load anything into this window it s a good idea to click the H at the top right of the waveform window This shows you the entire waveform and can return you to this display quickly following compression expansion Once the waveform is drawn listen to the entire utterance by clicking PLAY There are two additional options for listening to portions of the utterance including PLAY SEL selected and PLAY SEG seg ment The firs
11. stimulus required the client to monitor his respira tory support for speech and to end each breath group while still slightly above resting expiratory level That is when he finished the last word prior to an inspiration relaxation of the respiratory system would produce passive exhalation thus a small cushion of air The stimulus was constructed by recording the client using this strategy and then extracting a segment of passive exhalation from the record ing using the CUT and SAVE tools in Stimulus Generator Try your hand at this by clicking on LOAD A and finding the item SGSMP Stimulus Generator Sample in the Demo Stimuli folder You will STIMULUS GENERATOR Clinician first presents stimulus PLAY and then clicks RESP to indicate client s response if appropriate Stimuli are loaded or recorded into Windows A B C and D Stimuli can also be saved Running tally of each S and R is noted under each window A B C D 2 ew Client s um mise mem nee yeu STH cpu i Peay emer Prete fee oe icki name is entered mae Erste ee Clicking on SHOW under TALLY produces this summary of stimuli and responses which can be printed PT spooled or saved STIMULUS PLAY The name of each stimulus appears under each window This window 4 can be used to create new stimuli using the Selection Cut Copy Paste and A
12. LEO PO RATE RANGE CALCULATOR DM E E TE TF mi T o 356 211 106 1422 1778 213 11 4267 4622 4978 5333 msec LG FHr D RDO RATE on cient Bos OFF SOUNDSCOPE RECORD pLay sav SAY TO JNL SPEECH RATE TYPE CHANGE FO PARAMETERS stow ta ta ta Ser nec PLAY sau seL __sHow snl SEG ANALYZE EDIT sav am ccn enn ANALYZE MANUAL LOAD PLAY ccn enn SPEECH SEL SEGMENT SPEECH MAX SvLLABIC BATE 2 03 DURATION se 3 942 ses 4 143 APPLICATIONS LEO PO SPEECH RECONSTRUCTOR DM Lomo play _ nee 5 gt 55 SELECTED sew n paste me c cieee Play seo Af_s si gt so CUT __ noenaLize Save e zem prec oe INSTRUMENTS DESIGNED AND Zit 1067 1422 1728 2133 2489 2844 3200 3556 3911 4267 4622 4978 5353 msec Rebecca Leonard Ph D University of California Davis and Tito Poza M S Poza Consulting Services Menlo Park CA 708 1063 1417 1771 2125 2480 2834 3188 3542 3897 4251 4505 4959 5314 mSec 214 354 msec Div LEO PO STIMULUS GENERATOR DM Gwi SET REC LORD SAVE A cushion B eues on TALLY REC SELECTION _ SAVE a B A B A B PLAY Rese CLEAR Ta CLIENT C morph p Spool cory Prav f
13. ates the calculation of the diadochokinetic rate of the speech bounded by the two vertical marker lines in the waveform lower window The method used to calculate the rate is determined by the MANUAL AUTOMATIC control See Diado Function 6 When AUTOMATIC is selected as shown in the figure the rate is determined using a computer calculated syllabic count 22 DIADO FUNCTIONS CONT When MANUAL is selected the analysis relies upon a user entered syllabic count to determine diado rate The user can monitor the validity of the count by observing the correlation between the peaks of the blue SYL wave with the appropriate peaks of the red UTT wave Regardless of calculation method the result of the analysis is shown in the SYLLABIC RATE display See Diado Function 4 By clicking on SAV TO JNL See Common Function 14 SAV TO JNL the syllabic rate the utterance type along with the date time and the client s name can be recorded to the Journal for later archiving to disk 4 SYLLABIC RATE This is a display that shows the results of the diadochokinetic analysis See Diado Function 3 ANALYZE in units of syllables sec 5 UTT TYPE This display allows the user to document which of four specific utterances is to be analyzed The setting operates as a four way toggle It changes each time it s clicked cycling through pa pa ta ta ta ka ka ka and pa ta ka This setting is transferred to the logging journal along w
14. ay Selection This control allows the user to play the selected speech segment SAV UTT This control allows the user to save the speech in the wave called UTT as a file on the computer s disk drive SAV SEL Save Selection This control allows the user to save the selected speech segment as a file on the computer s disk drive SAV JNL Save Journal This control allows the user to save the data in the Journal as a file on the computer s disk drive The user may first wish to see the Journal See Common Function 15 SHOW JNL before saving it See also Common Function 14 SAV TO JNL The contents of the Journal are cleared after having been saved to disk 24 COMMON FUNCTIONS CONT 13 14 15 16 17 18 19 20 SEL Selection Duration This display shows the user the duration of the selected portion of the wave UTT It is loaded with a new value whenever the DURATION See Common Function 5 DURATION control is clicked If there is no selection when DURATION is clicked its value will be zero SAV TO JNL Save to Journal This control allows the user to save the data in the LOG window See Common Function 20 LOG to the Journal If the Journal already contains previously transferred data the new data are appended to it The user may see the current contents of the Journal at any time Gee Common Function 15 SHOW JNL SHOW JNL Show Journal This control allows the user to see the current contents of the Jou
15. bably skip over after reading about them once The manual is designed to be used as you are working with the computer screen in front of you There are some graphics in the manual which serve as overviews of the kinds of things each instru ment can do and of the controls needed to use them However the main graphic is the computer screen the illustrations in the manual are intended to serve primarily as references Once you have selected Rate Range Calculator and see it appear on the monitor the purpose of the instrument and how to use it will be explained in the text Just read along and try the operations de scribed It should be possible to use each of the demo instruments by simply reading the text perhaps occasionally referring to the graphics for additional information You will be able to load any of the demo stimuli into any of the instruments but you won t be able to record into SoundScope from other sources Nor will you be able to open up materials you ve manipulated and saved The three instruments demonstrated here were designed and created using the basic instrument design tools which are a part of SoundScope We believe they are a good illustration of the kinds of innovations in clinical instrumentation which are becoming more and more possible with contemporary computer technology As you use them you will hopefully think about other kinds of tools which might facilitate your own clinical experiences The instrume
16. ble a closer approximation to the FO values actually produced by the speaker Always keep in mind however that whenever you redraw any portion of the pitch trace your data are now an approximation of the actual track That is to say any statistics you calculate probably reflect the truth more closely than before you redrew the trace but they are not the result of a pitch tracking calculation An intermediate approach to correcting obvious pitch tracking mistakes is to use the NAN function Common Function 6 to eliminate obvious pitch tracking mistakes The name of this control stands for not a number and it allows you to make portions of the FO trace invisible to the statistics calculations It only works in the DRAW mode of the EDIT control Pitch Function 9 and involves a two step operation First click on the NAN button This will bring up a dialog box asking you to select a portion of the FO plot that you want to transform into values of NAN After clicking OK on the dialog box the cursor once again becomes an arrow and you should select with press drag release the portion of the FO trace that contains the outliers or obviously bad data Now click once more on the NAN button and the selected portion of the FO trace will disappear and the cursor will revert to a pencil As with the redraw clicking on ANALYZE Pitch Function 3 while still in the DRAW mode will recalculate the statistics on the modified data You may want t
17. culation worked correctly by counting the extended syllable as just one and not two syllables If it had erred however you could try another rate selection or alternatively a MANUAL calculation A MANUAL calculation is initiated by first selecting MANUAL on the MANUAL AUTOMATIC control Diado Function 6 Clicking on ANALYZE Diado Function 3 in the MANUAL mode bypasses the computer s syllable counting algorithm Determine how many syllables the segment contains by playing it and by counting the appropriate peaks in the waveform once again make sure the segment includes only those productions you want interpreted as syllables When you re satisfied click on ANALYZE Diado Function 3 A dialog box will appear asking you to enter the number of syllables which should be included in the analysis Type in the appropriate number and click OK The calculated number of syllables second will again appear in the SYLLABIC RATE window Diado Function 4 If this value is acceptable you may want to store it in the jour nal SAV TO JNL Common Function 14 You can also save the entire journal with SAV JNL but don t save it until you ve included everything in it you want The journal can be examined at any time by clicking SHOW JNL Common Function 15 The information you entered and the repetition rate for pa pa pa will be shown When you have completed all of your analyses and stored them in the journal you can select PRNJ Commo
18. de to more closely approximate normal Reconstructed segments can also serve as stimuli in tasks requiring the child to make judgments about reconstructed versus original productions In our experience even young speakers have become intrigued with cutting and pasting their own speech samples and then playing them Speakers who are mature enough to use the computer themselves to perform various operations may also engage in a lot of sound practice aloud as they manually manipulate their own utterances Clinicians will also find that the AMP feature may be particularly helpful in enhancing sounds which are otherwise difficult to attend to in speech contexts such as the weak f in the first example From the clinician s perspective another advantage of Reconstructor is its potential as a diagnostic tool For example in a speaker who presents with a number of sounds in error prioritizing the order in which they should be considered in therapy can sometimes be diffi cult Using the copy paste tools of SoundScope the clinician may be able to try out some corrections a priori in an effort to estimate their effect on the speaker s intelligibility There are of course many other criteria which need to be considered but being able to forecast even in a small way the possible effects of a particular correction or degree of improvement is potentially quite powerful Experimenta tion with Reconstructor will also provide the student
19. difficulty involved in FO analysis 10 If there appear to be areas in the FO trace where glitches of some type occurred outlying points far removed in frequency from the rest of the trace or an area where no trace is present for example you may want to do some manual editing of the trace Click on the EDIT button Pitch Function 9 to the right of ANALYZE until it becomes DRAW The cursor will turn into a pencil which can be used to manually eliminate the spurious or blank portions of the trace by means of a press drag release operation The idea here is to try to smooth out these areas by drawing lines connecting stable portions of the FO plot After the press drag release the redrawn trace will reflect the results of your edit Click again on ANALYZE Pitch Function 3 before you click out of DRAW back to LOG or EDIT and the pitch statistics will reflect the changes you made by redrawing the plot IMPORTANT If you change DRAW back to EDIT or LOG and click on ANALYZE the pitch plot will be recalcu lated and your redrawn pitch trace will be replaced Some of the statistics may not be very different after a redraw to correct glitches but the MAX and MIN will often show a signifi cant change if your redraw corrected a substantial outlier More importantly if you re just trying to fill in the questionable areas by following the progression of the plot as reasonably and carefully as possible the edit should ena
20. e recording of the stimuli and to construct exercises for the client to practice away from the clinic Our client for example used a Walkman recorder with a small 15 headset to receive the stimuli during monologue and reading exer cises and even in conversations A number of variables were ma nipulated during these exercises including the number of stimuli presented and the frequency of their occurrence over the duration of the tape If you try this it seems to work best if the stimuli from SoundScope are recorded from your system s audio out onto a good quality tape recorder at a fairly high amplitude This should enable the client to turn down the playback level of the portable cassette to a very low if not quite subliminal level making the loudness of the stimuli unobtrusive and minimizing any tape hiss or other back ground noise A final note our discussion has focused on the use of this instru ment with a fluency client but Stimulus Generator lends itself to other types of clients with voice articulation and perhaps even language disorders as well 16 SPEECH RECONSTRUCTOR The Speech Reconstructor tool represents one of the most novel features of SoundScope s clinical capabilities A long standing problem in the remediation of articulation impairments is what constitutes an appropriate accurate model of a sound which is in error particularly when the speech sound in error is part of a child s phonetic r
21. ed to fluency improved articulatory patterns voicing characteristics etc to travel with and be retained by the client in nonclinical situations Indeed it is the client s dem onstration of this facility which signals a successful treatment out come Stimulus Generator is based on the premise which seems reasonable to its developers at least that the ideal stimulus i e an event which prompts a client to utilize a strategy or target skill acquired through therapy is one which is unique to the client not to the clini cal environment and is therefore always with the client Beyond this it would be helpful if the stimulus always occurred during speech Obvious candidates are properties of the client s speech output itself or events which co occur with speech Examples of such stimuli are presented on the Demo Stimuli 2 folder The particular stimuli included on the demo disk were actually used with an adult fluency client and were tailored specifically to his needs Over a period of several months this client had developed a number of strategies which promoted his fluency and which in fact made the prospect of disfluency in his speech negligible Each strategy had been worked on individually until the client was adept at using it in the treatment room and ready to implement it in a struc tured manner in his daily activities If you haven t yet done so open Stimulus Generator To do this go 12 back to File in the
22. emember the order of your analyses Incidentally if you re SEGMENT pa pa pa with the rate set to SLOW all of the syllables will be included in the segment Look at the MANUAL AUTO control Diado Function 6 and note whether the window says AUTO or MANUAL For this first ex ample set it to AUTO and then click ANALYZE Diado Function 3 The number of syllables produced per second will appear in the The references in parentheses refer to the descriptions of the functions found in Appendix A pages 22 27 SYLLABIC RATE window Diado Function 4 Before you accept RATE RANGE s calculation you must first check to see which peaks the instrument included in its analysis On your monitor note the blue vertical lines in the waveform window below the red speech signal colors will be black if you don t have a color monitor Each blue spike denotes a production which was inter preted by CALCULATOR as a syllable and thus included in its calculation If a dysarthric speaker has added some extraneous utterances or if two syllables are extremely close together CALCU LATOR may make errors including something in the analysis that shouldn t be or indicating only one syllable when there were really two for example If the blue lines match up with the waveform and with what your ears tell you then the rate calculated is probably fine In this example in spite of the fact that the third syllable is inappro priately long the automatic cal
23. epertoire The problem of course is that the adult clinician typically uses his or her own production of the correct sound as a model of how the sound should be produced Obviously many features of this accurate version of the sound differ from the child s production not just place or manner of production for example but also features such as fundamental frequency formants and other spectral or temporal characteristics The child s task is made more formidable because at some level he must decide which of these features he is actually being asked to modify Speech Reconstructor can be used to extract certain features of the child s own productions for use in reconstructed utterances That is if a sound can be produced accurately in isolation or in particular contexts it can be easily copied and used to replace an inaccurate production in another context The reconstructed utterance can then serve as a model or stimulus in a variety of listening tasks From the File menu at the top of the screen select Open and find Speech Reconstructor When you ve opened it notice that your computer screen contains two spectrogram windows each above a corresponding oscillographic display or the waveform window of the speech utterance being examined The figure on page 18 presents an overview of Reconstructor s functions A more detailed explana tion of each of the functions on the Reconstructor control panel can be found in
24. es the vowel become more neutralized more schwa like formants equally spaced with time Are the syllables produced at equal intervals and are they all about equal in duration Separate analyses of beginning and ending segments may be revealing here A loss of If you own SoundScope you can open a saved Journal from the menu with Journal gt Load Text gt Journal then Edit gt Show gt Journal temporal or spectral integrity over time may reflect fatigue and provide clues to utterance lengths which may be optimal for the individual while more generalized irregularities may provide insights into the nature of the speaker s motor disorder For comparison return to LOAD SPEECH Common Function 4 and click TA TA TA DYS Remember to click on H when the waveform has loaded This time draw the spectrogram immediately by pressing the Cal button Auxiliary Function 1 in the right margin of the top window You may want to change the setup of the spectrogram for example to use a different analysis filter or to expand a portion of the frequency range Click on Set Auxiliary Function 2 and look at the parameters you can change Some Set options are available by clicking on the boxes others require you to type in values Try lowering the range of frequencies displayed from 8 KHz to about 5 kHz Click OK and redraw the spectrogram click Cal Are the formants easier to evaluate If consonant characteristics are
25. gram msec Div is the same The utterances will be loaded into the waveform windows below each spectrogram Click the button labeled Cal on the right margin of the window to draw the spectrogram You can click on the button labeled Set also on the right and change various characteristics of the spectrographic display For this particular exercise a filter of 300 Hz and a frequency range of 7 8 kHz is appropriate Remember you can also type in the names of the utterances under each spectrogram window The display can be expanded or compressed temporally at any time by clicking the arrows below and to the right of the waveform display msec Div Anytime you do this the spectrogram will be erased and you will need to redraw Cal it Once you have the utterances placed experiment with the PLAY A Reconst Function 6 and PLAY B Reconst Function 7 features to hear what you ve loaded SPEECH RECONSTRUCTOR Utterances are recorded or loaded into window A top or B bottom Elements from utterances in window A or B which have been selected can be Operations below can be performed on segmented bounded ur Segment A Segment B or Selected played with these tools Y PLAY ASB R SL SG INTERURL SELECTED SRUE R PASTE A gt B REC A LORD al PLAY PLAY sec A e si gt sco CUT __ _ __ ize snuE B REC e LOAD e PLAY e DUR 0 808 PASTE B gt A 4 loegmen
26. gt A Reconst Function 25 The f will be replaced with the segment of correct s you have isolated in the bottom display Play the upper utterance PLAY A and judge its naturalness You may need to do some fine tuning editing but probably not very much Re draw the spectrogram and check the visual results of your reconstruction Compare the corrected utterance now saw with the original faw in the lower display If you would like to try another reconstruction task click on LOAD A and find kikenz chickens in the demo stimuli When this utterance is loaded click on LOAD B and find ch isolation In this example the child is substituting k ch in contexts but is able to produce a reasonable ch with a schwa vowel Try your hand at correcting the child s production of kikenz in the upper display Copy paste functions can also be performed within a display upper or lower by using COPY and PASTE controls under the interval bar Try these as well As noted previously Speech Reconstructor exploits one of SoundScope s most interesting features The ability to isolate certain 20 elements of a speaker s sound repertoire and add them to or substi tute them for other elements provides some exceptional opportuni ties It affords a speaker with impaired articulation a chance to hear from a reconstruction of his own productions how he or she would sound if a particular error were corrected or were ma
27. h of 125 Hz 8 ms period there would always be 4 estimates for each pitch period One thing to watch out for with this parameter is to make sure you don t make it bigger than the smallest pitch period you expect to encounter or you ll lose information 2 ms is usually safe as it allows a 500 Hz FO with no loss of information The Re ject all peaks option allows you to make CALCULATOR more or less tolerant of samples it includes for analysis That is if it s set to 10 it will include for analysis only those pitch periods which vary no more than 1046 from adjacent periods The Reject feature can be particularly useful when you re trying to analyze a voice that contains significant aperiodicity or instability in frequency and amplitude components However you should recognize that making the analysis scheme more tolerant may also compromise the accuracy of results obtained to some extent The analysis of FO particularly in dysphonic voices is a complex and perilous task for any analysis scheme not just SoundScope When you re in doubt about data that you re obtaining remember that you can do a narrow band analysis or even expand a waveform to evaluate individual periods to help you decide if there is sufficient periodicity in the sample to get at least an estimate of FO In fact it s a good idea to do these other types of analyses regardless of the voice sample they may help you to better understand and appreciate the
28. his manual Once you ve done this you will notice a new SoundScope Clinical Manual f containing the SoundScope program the three clinical instruments available for demonstration Rate Range Calculator Stimulus Generator and Speech Reconstructor and two folders named Demo Stimuli 1 and Demo Stimuli 2 The stimuli in the first folder should be used with Rate Range Calculator the second folder contains stimuli for use with Stimulus Generator and Speech Reconstructor The program instruments and demo stimuli are the items you will need to be concerned with for the demonstration You must first open up the SoundScope application itself double click on it When you see a blank screen with a menubar at the top press the mouse on the File menu at the top left and select Open A dialog box will open up with a list of things you can open further Items listed include the three demo instruments and the two folders of demo stimuli The manual begins with an explanation of Rate Range Calculator and it is probably best to start with this instrument Highlight Rate Range Calculator with the cursor and click on Open There is a more detailed description of the functions and controls of this instru ment than in the material for the other two instruments The idea is that once you ve used these the others will be similar and familiar Some operations are described in the text for each instrument mak ing for some redundancies which you can pro
29. ime The active wave is the one whose name is highlighted in the left margin of the window In the DRAW mode the cursor changes from an arrow to a pencil and can be used to change the values of a wave by drawing over its current trace when it is the active wave in the window The wave is changed by pointing the tip of the pencil cursor at the desired place in the wave and doing a click and drag operation with the mouse When in the DRAW mode the NAN function See Common Function 6 NAN may be used to make portions of a wave invisible to analysis or logging functions A function See Pitch Function 49 EDIT similar to this one is found in the Pitch Instrument for altering pitch data 23 COMMON FUNCTIONS 1 10 11 12 CLIENT This control allows the user to enter the name of the client or subject whose speech is being analyzed This name will appear along with the current date and time as a header to any data logged to a journal RECORD This control allows the user to record speech into the wave named UTT The duration of the recorded speech sample will be determined by the current record settings See Common Function 3 SET REC LEVELS SET REC LEVELS This control allows the user to set the duration of the recording initiated by the RECORD function See Common Function 2 RECORD Record level adjustment and other hardware controls are also available using this function LOAD SPEECH This control allows the user to load
30. ith the calculated syllabic rate when the SAV TO JNL See Common Function 13 SAV TO JNL function is actuated Its only purpose is to provide documentation of which utterance was analyzed and it has no effect on the diadochokinetic analysis algorithm 6 MANUAL AUTOMATIC This control allows the user to determine the way the diadochokinetic rate is calculated See Diado Function 43 ANALYZE This setting is a two way toggle that alternates between MANUAL and AUTOMATIC when clicked When AUTOMATIC is selected as shown in the figure the rate is determined using a computer calculated syllabic count When MANUAL is selected the user is prompted to enter the number of syllables within the segment boundaries and this number is used by the program to determine diado rate 7 ON OFF This control along with the corresponding one for Pitch Analysis See Pitch Function 1 allows the user to choose between Diadochokinetic and Spectral Analysis or Pitch Analysis This setting takes on the values OFF and ON When it is in the OFF position as shown in the Figure none of the Diado Controls are functional 8 EDIT This control allows the user to change the cursor from its normal function EDIT to the DRAW function This setting is a two way toggle that alternates between EDIT and DRAW when clicked In the EDIT mode the cursor can as usual be used to select portions of the active wave in a window Only one wave may be active in a window at one t
31. ity of alternated syllables LPC analysis of formants is also available Common functions DIADO functions for Rate amp Range ON RATE RANGE CALC 6 14 GWI L Spectrogram User can es settings can be altered with these buttons timate speaker s rate to improve 1 621 4169 4516 4863 5211 accuracy of LG ud lls RATE I ON CLIENT rj L06 _orF PITCH ANALYSIS OFF Tt RATE CALCU SPEECH RATE a TYPE necorp_ _ptay wrr sav ro um CHANGE FO PARAMETERS FO functions LATOR and can te tate fevers SEG a oa ANALYZE EDIT OFF when ANALYZE AUTO zs reselect speecu SEL SEGMENT SPEECH CE using DIADO p S 7 15 syllable produced LL MX set 0 01065 SEG 1 238 E mx ve rom __ STD STD pu 7 166 7 16 Peat H assures that Analysis of all recorded DIADO RATE information is performed in sai TIME M mode Appears 347 Pea gate as syllables per second Expand or compress Waveform of syllable productions is placed in this window Syllables can waveform be automatically or manually segmented to include those desired for analysis Blue lines at bottom of window indicate energy concentrations SoundScope interprets as syllables Diado Rate Let s experiment first with the DIADO RATE instrument The task here of course is to determine how many syllables the speaker is ca
32. ly user selects which one by clicking Opens dialog box to allow user to load a stored speech utterance into waveform A B C or D Opens dialog box to allow user to save the utterance in waveform A B C or D to disk Initiates playback PLAY of speech utterance in waveform A B C or D RESP allows user to log client subject s appropriate response to stimulus presented in PLAY A B C or D Indicates all operations related to Tally Opens Journal and allows user to observe number of times each stimulus A B C and D has been presented number of times client subject has responded appropriately to each stimulus and percentage of times client subject has responded appropriately to each stimulus Opens dialog box to allow user to save data in Tally Clears tally Saves current tally and allows user to begin new trial or run Allows user to print contents of Tally individually or as totals and allows user to see entire contents in Tally Indicates all operations related to Selection Cuts an interval selected out of a speech utterance and copies it to a clipboard that corresponds in duration to the interval cut Copies an interval selected out of a speech utterance and stores it in a clipboard Pastes an interval cut or copied into the entry point in an utterance where the cross hairs are currently located Initiates playback of the selected portion of an utterance Opens dialog box to allow user to amplify selected
33. may want to check the FO trace once more One easy way to see if any voiced portions of the speech waveform have not been analyzed for pitch is to examine the green or black trace in the lower window above the red or black waveform trace This trace labeled _Pea is zero everywhere except where a pitch peak was found by the pitch tracking program By expanding the waveform you can if you wish zoom in on any problem areas and see what may have caused it If you expand the waveform you will also see the green or black band become a series of vertical lines each of which indicates a time at which the pitch tracking program found the start of a new pitch period Expansion will further reveal that the pitch track at the top of the display is actually a series of red or black dots The time between consecutive dots is determined by the Frame advance setting in the FO PARAMETERS dialog box described below If there appear to be unanalyzed areas at the beginning or end of the waveform you may need to adjust the range parameters accordingly You might also want to experiment with FO range settings to see how the analysis varies Going back into CHANGE FO PARAMETERS Pitch Function 2 reveals two other variables which can be manipu lated in performing analyses Frame advance determines the sample rate of the FO plot If it s set at 2 ms the FO wave will have 500 estimates for each second of speech If the speech has a constant pitc
34. menubar and select Open The figure pre sented on the next page is an overview of the Stimulus Generator instrument A more detailed description of the functions and controls of this instrument is presented in Appendix B on page 28 If you were going to record your own stimuli you could first set up the Record parameters with SET REC Stim Gen Function 1 and then RECORD Stim Gen Function 4 into the A B C or D win dow For now find the LOAD Stim Gen Function 5 buttons at the top of the display Click on LOAD A and open Demo Stimuli 2 For purposes of their use with Stimulus Generator the four strategies previously noted are referred to here as Cushion of Air Eyes On Morph and Stabilizer Load the first one into A and the other three stimuli into the B C and D windows respectively Click on the H in each window Also click on CLIENT and enter some identifi cation As you load each window the name of the stimulus representing each strategy will be placed over the appropriate set of PLAY and RE SPOND buttons Stim Gen Function 7 in the middle of the screen at the top When you ve loaded all four play each one of them We won t elaborate on therapy activities associated with the development of each of the strategies represented by the four stimuli However the stimuli do bear a close correspondence to the strategies For example play the first stimulus cushion of air The strategy represented by this
35. method allows you to keep track of the material you ve just isolated and lets you select new material for some other purpose At this point you may want to try typing some text into one of the text windows described earlier With this introduction to Speech Reconstructor it s time to try an actual reconstruction In the example provided the child is produc ing f s in saw but approximating appropriate s s in pencil and 19 house In the upper display select and isolate the f s in saw Move the vertical lines to delineate the exact portion you want to isolate Note the duration of the segment by checking the DURA TION Reconst Function 20 indicator remember you need to change the INTERVAL indicator bar to read SEGMENT A In the bottom display isolate a portion of the s from pencil or house preferably one that is similar in duration to the f When you have the two sounds isolated play each one of them several times and compare them As is apparent on the spectrogram f is weak before amplification and is widely dispersed along the frequency range Typically s is more intense and contains little energy below 4 kHz This three year old s s productions in pencil and house are perhaps mildly distorted but they do represent an improvement over the f s in saw When you re satisfied with the segments you have selected go to the upper right of the control panel and press PASTE B
36. mp stabilize tools in the upper right of Stimulus Generator Spooling Tally allows data to be saved as separate trials which can then become eyes on session by scrolling through SHOW or printing PRNT Tally 13 hear an example of passive exhalation following the breath group Isolate and save it using the tools in the upper right surrounding SELECTION Stim Gen Function 14 which subsumes Stim Gen Functions s 15 22 the operations listed around this term are performed on portions of the recording which have been selected or highlighted in green by pressing and holding the mouse button and dragging the mouse along the waveform You can use this feature to identify the passive exhalation in the sample you ve loaded just highlight something and click on PLAY Stim Gen Function 18 When you ve isolated the portion you want you can SAVE Stim Gen Function 22 it Now play window B containing the stimulus eyes on This stimu lus served as a cue to the client to maintain or restore eye contact with a listener It was constructed by extracting portions of two p s from something the client had said This particular stimulus was felt by the client to be especially appropriate the two rapid clicks representing two eyes and the abrupt transient character of the noise serving to quickly grab his attention Clicking on morph reveals an exaggerated production of CV syllable ka In therapy a grea
37. n Function 18 to produce a hard copy The utterance or a selected portion of the utterance can also be saved to disk by clicking on the appropriate save control Common Functions 10 SAV UTT and 11 SAV SEL Clicking on one of these controls will bring up a dialog box asking you to name the information you want to save Scroll through the list box above the name to find the folder where you d like to store your data Double click on it type in a name for your data and click SAVE You can also SAV JNL Common Function 12 to disk but this will clear your working journal so wait until you have everything in it you want Repetition rate is one measure which can be elicited from a diadochokinetic task But clinicians who consider only rate may be overlooking some very useful information RATE RANGE CALCULATOR S design facilitates a more comprehensive assess ment of syllable repetition tasks Let s consider some examples starting with the utterance you ve already loaded Go to the window at the top of the screen display look for Cal Auxiliary Function 1 in the right margin and click it A spectro gram of the pa pa pa will be drawn Examine the uniformity of the syllables produced by this speaker Is there a clear distinction be tween consonant and vowel Are there instances where the conso nant is very weak unaspirated voiced or even missing Does the formant structure for a appear appropriate across all the syllables or do
38. n isolation You will also note that the child is substituting f s in this word If f is difficult to hear use your mouse to select only the fricative at the beginning of the word and then go to the controls and find the INTERVAL indicator button Reconst Function 14 This button will read INTERVAL followed by either SEGMENT A SEG MENT B or SELECTED You can change from one setting to the other by clicking on the bar When you have INTERVAL SE LECTED showing push on the AMP Reconst Function 17 button located just below the INTERVAL bar A dialog box will open requesting you to enter an amplification factor 100 means no change It s usually best to start with small values for example 120 or a gain of 1 2 Type this number in the box and click OK The portion of the utterance you have selected in this case f will now be amplified You can hear the results immediately by clicking on PLAY A or PLAY SL However you will have to click on Cal again to redraw the spectrogram and see the results When you ve redrawn the f compare it to the one in the lower display The second way to isolate a particular portion of an utterance in volves using the left and right vertical lines called segment markers present in both sets of displays If you hold down the Option key you will find that when the mouse cursor is over a waveform or spectrogram window its icon changes from a diagonally pointing arrow to
39. n of the wave UTT bounded by the two vertical markers It is reset to a new value whenever the DURATION See Common Function 5 DURATION control is clicked LOG Logging Window This display holds any data that is recorded using one of the logging functions See Diado Function 1 LG FMT and Pitch Function 9 EDIT When desired the user may save the data to the Journal See Common Function 14 SAV TO JNL which also clears the window This window may also be edited manually by the user 25 PITCH FUNCTIONS 1 ON OFF This control along with the corresponding one for Diado Analysis See Diado Function 7 allows the user to choose between Pitch Analysis or Diadochokinetic and Spectral Analysis This setting takes on the values OFF and ON When it is in the OFF position none of the Pitch Controls are functional 2 CHANGE F0 PARAMETERS This control allows the user to modify three of the parameters that affect the pitch tracking algorithm The Acceptable Range should be adjusted to minimize outliers The effect of the frame advance and allowable variation in adjacent pitch periods controls is more complex and the user may wish to experiment with these parameters before collecting data 3 ANALYZE This control initiates the calculation of pitch rate data for the speech in the wave UTT bounded by the two vertical marker lines in the waveform lower window With the EDIT control See Pitch Function 9 EDIT set to EDIT the pitch track is
40. nd LOG when clicked In the EDIT mode the cursor can as usual be used to select portions of the active wave in a window Only one wave may be active in a window at one time The active wave is the one whose name is highlighted in the left margin of the window In the DRAW mode the cursor changes from an arrow to a pencil i and can be used to change the values of a wave by drawing over its current trace when it is the active wave in the window Ihe wave is changed by pointing the tip of the pencil cursor at the desired place in the wave and doing a click and drag operation with the mouse When in the DRAW mode the NAN function See Common Function 6 NAN may be used to make portions of a wave invisible to analysis or logging functions A Diado Function See Diado Function 8 EDIT similar to this 26 PITCH FUNCTIONS CONT one may be used for altering formant data In the LOG mode a vertical marker line will appear at a 0 5 seconds in the pitch analysis window and the cursor icon will become the symbol which when lined up the marker will allow the user to move the marker to a new location by executing a click and drag operation When the marker is moved the value of the pitch at the new position will be recorded in the LOG window See Common Function 20 LOG If the logged values are to be preserved the LOG must be saved to the journal See Common Function 13 before this control is changed back to EDIT or the LOG windo
41. ne vibratory cycles Holes in plot can be edited 10 or eliminated 3320 msec 15 20 msec Div The first step in performing the range analysis is to set the FO param eters to make CALCULATOR s FO estimates as accurate as possible Click on CHANGE FO PARAMETERS Pitch Function 2 A dialog box will open allowing you to adjust some analysis param eters If you re looking for mean values from connected speech tasks you might want to check one of the boxes appropriate for an average adult female or male However if you want to assess range data the third option is especially useful This option allows you to enter a low and high FO value respectively which will determine the range of allowable frequency values Click on each box and type in your selections If you haven t had much experience estimating FO s by ear you may need to adjust the range and repeat ANALYZE Pitch Function 3 a couple of times When you ve set the FO parameters close the dialog box and click on ANALYZE Pitch Function 3 When CALCULATOR is finished you will see the red or black FO trace or melody plot in the upper display Values for mean FO and standard deviation duration of the segment analyzed and minimum and maximum values in the sample assessed will also be displayed Pitch Function s 4 5 6 7 and 8 As with DIADO RATE you may choose to save these data in the journal you ve created Before you do this however you
42. nts are fairly user friendly intuitive and even fun to use We think you ll enjoy them RATE RANGE CALCULATOR Rate Range Calculator is designed to be of particular use in the evaluation of a speaker with a neurogenic disorder It actually con sists of two separate instruments that share the same screen The figure presented on the next page is an overview of the Rate Range instrument A more detailed description of the functions and controls for each of the two instruments DIADO and PITCH as well as for those controls that are shared by both instruments COMMON is presented in Appendix A on pages 22 27 The Rate instrument can assist the clinician in assessing syllable repetition tasks also referred to as diadochokinetic tasks Beyond a simple calculation of rate syllables sec however the analyses performed on rapidly repeated utterances can provide the clinician with information regarding other articulatory parameters unique to the speaker including the temporal regularity of syllable pulses and the spectral integrity of utterances over one breath group The Range functions of Calculator are useful in evaluating prelimi nary fundamental frequency and durational characteristics of voicing The instrument is not intended for those situations in which a com prehensive voice evaluation is desired but rather for those instances when information regarding possible limitations on a speaker s ability to sustain sounds turn the voice
43. o expand the pitch trace in order to assure greater preci sion in selecting the portions of the trace to eliminate A final feature of the Range instrument can be exercised by clicking on the EDIT button Pitch Function 9 until it becomes LOG This activates CALCULATOR s log F0 to journal function As you did with the LG FMT function of DIADO RATE find the vertical marker that appears at 0 5 seconds in the upper display As with the LG FMT function if you line up the cursor on this marker and use a press drag release operation you can move the marker to a desired point in the FO trace When you release the mouse the FO value at the point of the vertical marker as well as the time will be entered in the journal These values can be stored in the journal SAV TO JNL Common Function 14 and eventually printed out PRNJ Common Function 17 if you like 11 STIMULUS GENERATOR Stimulus Generator is designed to facilitate stages of therapy in which generalization or carry over activities are critical This aspect of perhaps every behavioral therapy presents some unique challenges for both clinician and client The extension of skills acquired and used with facility in the clinical environment to the client s real world environment is the last and sometimes most diffi cult stage of the therapeutic process Ideally we would like those stimuli which in the context of therapy seem to elicit desired re sponses in the client relat
44. of speech science with some interesting and valuable lessons regarding the nature of speech particularly as it relates to coarticulatory phenom ena and the multiplicity of cues which a listener responds to in making decisions about what he s heard The influence of transitions and durational variables may become particularly apparent for example if sounds from very different environments are inter changed Some manipulations hopelessly violate the naturalness of speech others result in very natural sounding productions The instrument is a great educational tool even if your speech science background seems long ago and or far away you still have your ears and brain to help you decide if what you ve created with Speech Reconstructor is appropriate or not 21 APPENDIX A RATE RANGE CALCULATOR CONTROLS DIADOCHOKINETICRATE CONTROLS AND DISPLAYS COMMON TO BOTH INSTRUMENTS PITCH ANALYSIS CALCULATOR CALCULATOR LG FHT DIRDO RATE OFF CLIENT LOG on PITCH ANALYSIS geli i RECORD PLAY SAU sav TO UNL SPEECH RATE UTT CHANGE FO PARAMETERS WESS C UP em pre PLAY snu sev sHow ANALYZE SEG sdb one PRAY HALYZE Eia m E CE 171 D j A sive ojos sep pap epo nels vp eS EE ber Lee srh 75 ola coole ele Oo 9909000000000 0600 1 DIADOCHOKINETIC FUNCTIONS LG FMT Log Formants This control alow
45. ool in effecting the desirable treatment outcome referred to earlier That is elements of the speaker s own speech output eventu ally provide the stimulation necessary to implement specific thera peutic strategies in nontherapy situations a source of stimuli which being self generated is always with the speaker and always present during speech Initially of course the stimuli are presented by the clinician and the client is responding to this external intervention In time however and with frequent intense practice the same or similar stimuli which occur naturally in the client s speech may acquire the potential of the experimental stimuli and elicit the same desired responses In the example described here the four stimuli were all presented to the client during each therapy exercise How ever it should be recalled that this client was well into his treatment and had demonstrated facility with each strategy independently before he was asked to use combinations of them first in pairs then in groups of three and finally all four of them within a single con versational exercise With another client the clinician may want to use Stimulus Generator to present only a single stimulus perhaps at discrete intervals such as once in five sentences or at particular points in a reading passage or at certain time intervals Another possibility one which was beneficial with the adult fluency client referred to here is to make a tap
46. pable of producing per second To begin be sure that the button to the right of DIADO RATE Diado Function 7 is set to ON Click on CLIENT and type in a name this will now appear on the screen The RATE RANGE CALCULATOR can automatically segment the pa pa pa utterance for you if you click on SEGMENT SPEECH Common Function 18 If you re not happy with this segmentation however you can move the vertical lines as described above to bound the segment you want included in the analysis Try SEG MENT and observe where CALCULATOR places the segment markers The automatic segmentation may not include the last three syllables This is due to the break between the first 13 syllables and the last 3 One could argue that including the entire utterance in the segment would not give the appropriate rate for this speaker How ever if that is what is desired the right marker can be moved manu ally to include the last 3 syllables in the segment as well Now click under SPEECH RATE Diado Function 2 to choose a rate which you think characterizes the speaker s rate we ve selected SLOW for this speaker Go to UTTERANCE TYPE Diado Func tion 5 and find pa pa pa RATE RANGE CALCULATOR provides four options here including repetitions of the syllables pa ta or ka as well as the alternated pa ta ka These options are included so that any data which might be stored in the journal will be appropriately labeled and you won t have to r
47. portion of a speech utterance by some factor Calculates the duration of a selected interval 21 Shows duration of interval Calculated Opens dialog box to allow user to save the selected portion of an utterance to disk 28 APPENDIX C SPEECH RECONSTRUCTOR CONTROLS OOOO o a LOAD a a FER PLAY Age a sL gt so INTERVAL SELECTED SAUE A paste ADB pa PLAY SEG A B sL sG CUT ZE AWE REC e LOAD e PLAY e ees Bl piy a pun 0 808 Roe sL PASTE gt Initiates recording into waveform A according to current control settings Initiates recording into waveform B according to current control settings Opens dialog box to allow user to load a stored speech utterance into waveform A Opens dialog box to allow user to load a stored speech utterance into waveform B Opens dialog box to allow user to change the record settings e g time of recording Initiates playback of speech utterance in waveform A Initiates playback of speech utterance in waveform B Initiates playback of speech utterance in waveform A followed immediately by utterance in waveform B Initiates playback of that portion of the utterance in waveform A that lies within the two segment markers 10 Initiates playback of that portion of the utterance in waveform B that lies within the two segment markers 11 Moves the waveform A segment markers to coincide with the beginning and end of the
48. pparent in both the waveform and the spectrogram the syllables appear to have been produced with about the same intensity In short the ta repetitions reflect the precision and orderliness typical of normal speech in the adult You may want to go back to the previous examples of PA PA PA DYS and TA TA TA DYS and note if other differences are apparent between the normal and dysarthric speakers One further type of investigation possible with RATE CALCULA TOR is formant tracking using SoundScope s LPC analysis capabili ties This further scrutiny of diado productions might be desired when your initial impression is that the formant structure of the speaker s vowels changes over time on the syllable repetition tasks When you have one of the waves loaded go into the spectrogram Set box where you previously changed the range of frequencies displayed At the bottom of the dialog box that opens when you click on Set is an Options button Clicking on this reveals choices you can make regarding which formants you d like tracked The first two or three are usually sufficient Check the ones you d like displayed and then close the dialog box and redraw the spectrogram click Cal When the spectrogram redraws this time you will be presented with a message informing you that the LPC tracking operation is underway This may take a little while it s complicated process When the operation is completed you will see colored traces re
49. presenting the formants you selected for tracking These will be difficult to see if you don t have a color monitor sorry Remember that the lowest formant in frequency is always considered formant 1 the next lowest in frequency formant 2 and so on If the syllables weren t expanded in this initial analysis you may want to expand the time waveform and do the LPC analysis again This might make it easier to visualize the formant locations across the vowels You might also want to adjust the frequency range displayed to facilitate viewing the formants When you ve changed the settings try the LPC analysis again Do and F2 appear to remain stable across the syllables Are they uniform in time If the speaker has difficulty terminating voicing you may see a lot of transition activity in F1 and F2 as the vocal tract changes shape while voicing continues To the left of DIADO RATE on the control panel there is a button labeled LG FMT Diado Function 1 Clicking on this button activates CALCULATORS s log data to journal function You should now see three vertical black segment marker lines in the spectrographic display The one that appears at exactly 0 5 seconds activates the logging process Selecting the wrong marker may cause the spectro gram to disappear again use Cal to redraw Place the cursor now the icon directly over this marker and perform a press drag release to a steady state portion of the speaker s first vo
50. rnal When the viewing is complete the user can return to the instrument by either closing the Journal by clicking in the close box at the upper left hand corner of the Journal window or by clicking on any part of the instrument on the screen behind the Journal CLRJ Clear Journal This control allows the user to clear the contents of the Journal without having to save it first PRNJ Print Journal This control allows the user to print the contents of the Journal The contents of the Journal remain unchanged after printing SEGMENT SPEECH This control allows the user to automatically create a segment within the wave UTT by bounding a portion of the speech by two vertical markers This segment defines the speech interval that is to be processed by the various analysis functions See Diado Function 2 ANALYZE Common Function 8 PLAY SEG Common Function 19 SEG Pitch Function 3 ANALYZE Before initiating any analysis on this segment the user should always listen to the bounded speech by clicking on PLAY SEG If the interval has not been appropriately segmented the user can correct the segmentation by manually repositioning the vertical markers This is accomplished by holding the option key down which changes the cursor icon tod placing the cursor icon over one of the markers and performing a click and drag operation to move the marker to a new location SEG Segment Duration This display shows the user the duration of the portio
51. s the user to log formant values It only works when the Diado instrument is turned on The function is toggled on and off by clicking on it Clicking on it causes it to turn off if it is and to turn if it is off As shown above it is off When it is turned on a vertical marker line will appear at a 0 5 seconds in the spectrogram window and the cursor icon will become the symbol which when lined up the marker will allow the user to move the marker to a new location by executing a click and drag operation When the marker is moved the values of the formants at the new position will be recorded in the LOG window See Common Function 20 LOG If the logged values are to be preserved the LOG must be saved to the journal See Common Function 13 before this control is turned off as turning the Log Formant control off clears the LOG window Naturally if formants have not been calculated See Auxiliary Function 1 Cal prior to using this function their values will be logged as zero SPEECH RATE This control optimizes the diadochokinetic analysis algorithm on the basis of the relative syllabic rate of the talker The setting operates as a three way toggle It changes value each time it s clicked cycling through SLOW MODERATE FAST Clicking on this control does not cause an observable result it simply sets parameters for the ANALYZE See Diado Function 3 function ANALYZE This control initi
52. selected portion of waveform A 12 Moves the waveform B segment markers to coincide with the beginning and end of the selected portion of waveform B 13 Initiates playback of the selected portion of a speech utterance 14 Indicates the part of a speech utterance either Segment A Segment B or the Selected portion that will be the target of the five functions and one indicator located below Clicking on the bold title changes the name of the target interval 15 Cuts the interval indicated above out of the speech utterance and copies it to a clipboard that corresponds to the interval 16 Copies the interval indicated above to the clipboard that corresponds to that interval 17 Opens dialog box to allow user to amplify or attenuate the interval indicated above 100 leaves it unchanged 18 Pastes the interval currently on the clipboard to the entry point in the utterance that corresponds to the interval indicated above at the point where the crosshairs are currently located 19 Normalizes the interval indicated above 20 Shows the current duration of the interval indicated above 21 Opens dialog box to allow user to save the utterance in waveform A to disk 22 Opens dialog box to allow user to save the utterance in waveform B to disk 23 Opens dialog box to allow user to save the selected portion of an utterance to disk 24 Pastes segment A into the utterance of waveform B starting at the first segment marker for waveform B 25 Pastes segment B
53. sights into a speaker s capabilities These kinds of measures may be especially useful when the clinician is trying to determine what sort of respiratory and phonatory support for speech are available to a particular speaker In order to try out Range Calculator you will first need to turn on the PITCH ANALYSIS instrument by clicking the Pitch ON OFF button Pitch Function 1 When the program is ON the display at the top will change from a spectrogram window to a pitch plot window The controls in the middle of the control panel are common to both DIADO RATE and PITCH ANALYSIS so click on LOAD SPEECH Common Function 4 and find FO Range in the list of Demo Stimuli 1 Click on it and wait for it to be loaded into the waveform window at the bottom of screen Once it s loaded remember to click on the H at the top of the right hand margin of the waveform display so that you can see the entire waveform Then segment the portion of the waveform you want to analyze by using SEGMENT SPEECH Common Function 18 or by putting the segment markers around the part of the waveform you want to consider RANGE CALCULATOR RATE RANGE CALC 6 14 E 0 500 1000 1500 2000 2500 2000 ODO 4 500 msec LG n RATE orr criEnr i L06 on PITCH ANALYSIS RECORD PLAY sau urr sav ro umu SPEECH RATE UTT TYPE CHANGE F PARAMETERS SET rud PLAY sav SEL SHOW JNL J FAST ta ta ta LEVELS SEG ANALYZE ED sau JML cLRJ
54. t A The f s produced d can replace in saw can be Segment B reconstructed or B can using a more n MC replace A appropriate production of s extracted from the same speaker s production of D pencil 354 msec Div Weak fricatives in the productions can be selected or segmented and then amplified AMP to provide better feedback 18 or press PLAY A amp B Reconst Function 8 There are two ways of isolating particular portions of the utterances for further scrutiny First you can use the selecting feature com mon to all Macintosh applications If you are familiar with Macintosh word processing applications you will find that selecting portions of an utterance is essentially the same as selecting text For example locate the first word in the upper display containing wave form A Move the cursor to the beginning of that word press and hold the mouse button down and drag the mouse to the right As you move from left to right you will see that the part of the waveform you selected becomes highlighted Incidentally you may notice that the first click in a waveform window after clicking somewhere else does not start the selection process It takes one click to activate the window before you can start selecting When you ve highlighted saw move to the controls at the top of the display and click on PLAY SL Reconst Function 13 You will now be able to hear the word saw played i
55. t CALCU LATOR to interpret as syllables Again notice the repetition rate for ka as compared to those determined for pa and ta Any in sights regarding the integrity of anterior vs posterior tongue Try expanding the waveform with the arrow on the left side of the time window at the bottom right of the display Expanding the signal and then redrawing the spectrogram click Cal provides graphic evidence of the dipthongization apparent in the speaker s utterances Was this characteristic of the earlier productions as well Now click LOAD SPEECH find TA NORMAL load it and draw its spectrogram In some contrast to the utterances you ve just ana lyzed the normal speaker s syllable repetition rate is greater than seven syllables per second But as we ve noted there are many other features of diadochokinetic tasks which may provide insights into a speaker s articulatory finesse In this example consider the regularity of the syllable pulses produced in both temporal and spectral do mains Each syllable is of approximately the same duration and the intervals between them are also equal There is a clear demarcation between consonant and vowel and the spectral characteristics of the syllables appear homogeneous from start to finish The formant structure is appropriate for a and unlike the previous utterances there is little or no diphthongization of vowels Voicing does not continue from one syllable to the next and as a
56. t deal of attention had been focused on the client s manipulation of various types of utterances particu larly on transition elements within and between utterances For example the syllable ka can be produced with the k and the a almost completely coarticulated or co occurring But it can also be produced appropriately in many other ways by manipulating voice onset time the degree of aspiration on k the strength of tongue contact against palate for k and so on Attention to this strategy was helpful to the client s realization that he could control his own output could in effect orchestrate certain parameters of speech to his own liking without sacrificing intelligibility or content or fluency He selected the particular form of ka presented here as a stimulus which he felt would be a powerful cue to him to implement morph in conversational exercises 14 The final stimulus here referred to as stabilizer was extracted from an utterance containing s Stabilizers i e slight prolongations of sustainable sounds were used by the client for a dual purpose First they served as brief pause periods within connected speech Dur ing these periods the client did a quick check of his respiration was he exhaling smoothly and was he aware of gradually diminishing lung volume On another occasion he might use the stabilizer pause to focus briefly on eye contact and to re establish it if needed From the sample utterance
57. t in the text window in the lower right corner of the screen You should only use the SHOW function when a trial is complete since a totally new summary will be ap pended to the display each time you click on SHOW With other tools under TALLY this information can be saved to disk SAVE Stim Gen Function 10 or printed PRNT Stim Gen Function 13 and the log cleared CLEAR Stim Gen Function 11 for the next trial or task SPOOL Stim Gen Function 12 allows you to save multiple summaries in the journal This feature is nice if you want to conduct several trials with a client and then compare performances across all of them When you re finished with a session you can save the journal to disk SAVE However if you do this remember that all the current journal information including client identification will be erased in preparation for a new start ABORT Stim Gen Function 3 will also clear all existing data PRNT opens a dialog box pre senting several options for printing including the current tally all the tallies or all the tallies and a Grand Total summed across tallies of the results PRNT also offers a fourth option which enables you to see a summary of all tallies and totals on the monitor screen This is handy since it lets you view the contents of the journal containing tally information without having to print it out each time In our experience Stimulus Generator has proved to be a useful clinical t
58. t requires highlighting or selecting a portion of the waveform that you d like to hear Do this by first clicking on the label if it s not already highlighted in the left margin of the waveform window Then put the mouse cursor on some part of the waveform press and hold the mouse button down drag the mouse and cursor from left to right until you ve isolated what you want then release the mouse button The portion of the utterance you ve se lected should be highlighted Now try clicking PLAY SELECTED You can deselect by simply clicking again on the selected area Another way to isolate part of the utterance is to bound it with two markers vertical lines in the waveform window that create a segment containing the portion desired Each of the two vertical lines can be moved by 1 holding down the Option key on the keyboard which changes the cursor icon from an arrow to 2 lining up the vertical line of the cursor with the vertical line of the marker and 3 pressing and holding the mouse button while dragging the line to a new location If the marker seems hidden look for a dashed vertical line at the extreme right or left edge of the window which indicates that it is off the screen It can be brought into view by lining the cursor up exactly with the dashed line and carrying out the process described above RATE CALCULATOR Spectrogram of diado utterance permits examination of spectral and temporal regular
59. there is a need for precise control of the length of stimuli you want to construct Once the particular stimulus or stimuli have been constructed and loaded into one or more of Stimulus Generator s windows you can begin to use it In a real therapy situation you might want to click on ABORT Stim Gen Function 3 to clear all previous records and then click on CLIENT Stim Gen Function 2 to begin a new record The client would be presented with his or her stimuli through an earphone connected to the audio output of SoundScope or your computer While the client engages in conversation with the clinician or with others the clinician presents him her with stimuli The first time you click on PLAY Stim Gen Function 7 Stimulus Generator will ask you to enter a client name if you haven t already done so Notice that each time you click on PLAY Stim Gen Function 7 this presentation is recorded under the appropriate window as Played If the client responds to the stimulus with the appropriate strategy the clinician then clicks on the corresponding RESP button This information will also be logged under the appropriate window A running tally of the number of presentations of each stimulus and the number of appropriate responses to each will continue until the exercise is ended Clicking on SHOW Stim Gen Function 9 under TALLY Stim Gen Function 8 will produce a summary of the client s performance and display i
60. w will be cleared and the data lost AUXILIARY FUNCTIONS Pitch Analysis ES e Diado Analysis Configuration B Configuration 1 Cal Calculate This control allows the user to initiate either spectrographic analysis in the Diado configuration or pitch analysis in the Pitch configuration independent of the Diado and Pitch instruments described above When pitch analysis is carried out by clicking on Cal rather than on the ANALYZE control of the Pitch instrument no FO statistics are calculated When Cal is clicked on in the Diado Analysis Configuration the sound spectrogram of that portion of the wave UTT that is bounded by the two vertical markers is drawn in the window This is followed by a formant analysis that uses a peak picking algorithm on the calculated LPC spectra 2 Set Settings This control allows the user to adjust various settings that affect either the pitch analysis or the spectrographic analysis depending on which configuration is in effect 27 a ad ae 09 APPENDIX B STIMULUS GENERATOR CONTROLS ILLI LOAD SAVE SELECTION SAVE E e hav nese nese efef clo clo may nese pray nese Opens dialog box to allow user to change the record settings e g time Opens dialog box to allow user to enter client subject identification Opens dialog box to allow user to terminate current data run trial Initiates recording into waveform A B C or D respective
61. wel that is to a point where the LPC traces appear fairly flat When you release the mouse button you will see the formant values and the time of their occurrence appear in the LOG window Common Function 20 By repeating this press drag and release operation at selected spots in the various syllables you will record representative formant data across the speech sample When you ve finished you can scroll through the log to see if the quantitative data support your impressions of what was happening to the vowels over time If you want to save these data in addition to your summaries for diadochokinetic rates just repeat the SAV TO JNL Common Func tion 14 operation for printing now or later One caution once the LPC traces are displayed they stay on the screen when you redraw the spectrogram even if you go back into Set and remove them by turning them off in the Options win dow Since it takes a little time to do the LPC analysis you probably won t want to have this happen every time you redraw the spectro gram in particular when you re changing the range of frequencies displayed to look at consonant characteristics or experimenting with expanding the time scale In short the LPC analysis should be the last analysis you perform prior to loading a new utterance You might also want to turn off the LPC analysis when you finish with CALCU LATOR so that the next time you use it you won t forget and start with an LPC anal
62. ysis the first time you draw a spectrogram There is a control called NAN Common Function 6 that enables you to make portions or all of a formant trace invisible and this function can be used to get rid of formant traces without having to load a new utterance The description of the control found on page 24 Com mon Function 6 explains how to accomplish this f the spectrogram disappears just click Cal again Pitch Analysis The buttons on the right side of Rate Range Calculator control the instrument s fundamental frequency analysis functions See the figure on the next page for an overview of the screens produced with Pitch Analysis See Appendix A pages 26 27 for more detailed descriptions of the pitch functions The Range portion of CALCU LATOR was designed to assist the clinician in obtaining some pre liminary information about a client s voice production capabilities It is useful in assessing a speaker s fundamental frequency range determining mean FO s in connected speech and observing changes in FO associated with durational variables For example utterances which become progressively longer as Many men made millions Many men made millions on rainy days Many men made millions on rainy days in January and June may reveal interesting variability in FO as well as in voice quality Similarly such tasks comprised first of all voiced utterances and then voiced voiceless compo nents may also provide in

Clinical Applications Manual

Contents

Download Pdf Manuals

Related Search

Related Contents