BACHELOR THESIS
Contents
1. Lite o Lite Lite CJ N gorlunda CJ N gorlunda o N gorlunda Q Mycket O Mycket 0O Mycket V ldigt mycket v ldigt mycket a v ldigt mycket O V ldigt mycket 19 Questions after tasks Vad var ditt vergripande intryck av sv righetsgrad vid mus och tangentbordinteraktion o o o o o V ldigt svart Svart verkomligt Enkelt Valdigt enkelt Vad var ditt vergripande intryck av sv righetsgrad vid interaktion via kontrollytan o o o o o V ldigt svart Svart verkomligt Enkelt Valdigt enkelt Hur stor effektivitetsskillnad mellan de tv olika interaktionsmetoderna noterade du i arbetstl det o 0 o o o V ldigt liten Liten N gorlunda Stor v ldigt stor Vilken interaktionsmetod f redrog du o o o Mus tangentbord Kontrollyta ingendera Om du f redrog n gon av interaktionsmetoderna f reg ende fr ga motivera Andra tankar eller kommentarer kring interaktionen du vill dela med aig av 202. peaking EQ user interface for single channel corrective EQ tasks AES Convention paper 8855 Presented at the 134th AES Convention Rome 6 Jay Hodgson 2010 A field guide to equalisation and dynamics processing on rock and electronica records Popular Music 29 pp 283 297 doi 10 1017 S0261143010000085 7 Massenburg G 1972 Parametric Equalization Presented at AES 42nd Convention Los Angeles 8 NASA TLX Task load index Historical reference for NASA TLX Retrieved 2014 04 23 from http humansystems arc nasa gov groups tlx index html 9 Behringer BCR 2000 product description Retrieved 2014 04 23 from http www behringer com EN Products BCR2000 aspx 10 Apple Inc MainStage 2 user manual Retrieved 2014 04 7 from https documentation apple com en mainstage usermanual index html chapt er 1 26section 1 26tasks true 11 Waves Renaissance Equalizer Manual Retrieved 2014 04 23 from http www waves com 1lib pdf plugins renaissance equalizer pdf 18 Appendix Questions between each task Hur val lyckades du genomf ra det du blev tillfr gad O Valdigt daligt D ligt Hur frustrerad var du O V ldigt lite o Lite C Acceptabelt CJ N gorlunda Mycket Hur mycket tid spenderade du f r att genomf ra uppgiften O v ldigt lite Hur os ker var du o v ldigt lite Hur mycket jobb kr vdes f r att stadkomma den niv du presterade O Valdigt lite o
3. to watch the screen and at the same time changes the parameters could be an advantage At the same time the parameter information displayed by the knobs provides an important source of data and if the user can choose to attend to the information whenever it is necessary it creates a valuable tool for verification and reference The basic parameters of the equalizer are the frequency band q value and gain The q value is how wide frequency spectrum that will be affected and gain is the amount of boost or cut in the chosen frequency area There is also different type of filters where the most common is peak shelving and pass filters The pass filter low or high let all the frequencies above or below the set frequency pass and takes away the rest The peak filter creates a dip or peak in the frequency spectrum around the chosen frequency with a bell shaped curve which width is dependent on the q value The shelving filter low or high is designed to increase or reduce the frequencies below or above the set frequency with an equal gain The pass and shelving filter also has a slope depending on the filter construction or setting See Fig 1 6 7 Fig 1 Curve examples of a high pass filter a peak and low shelving filter with increased gain There are a number of typical trends in the ways people use the equalizer 6 In the workflow to improve the sound of single instruments in a mix it is common that engineers first cut away the unwante
4. you are selecting clicking and dragging visual knobs faders and buttons on the screen typing in numbers or use combined mouse and keyboard combination to get quick access to certain functions In that situation there could be bad precision struggle with zoom commands and knobs who has unnatural movement patterns for the mouse Because these often are substitutes for real analogue counterparts with higher precision and a natural tactile input it could be preferred to drag in the EQ curve which is more natural and effective with the mouse 4 Another problem to consider with visual displays is the fact that it can make the user want to draw a good looking EQ curve instead of listening to the results The user can shift the focus from listening to watching because of the limitation in human perception 2 5 The equalizing task The adjustments of the equalizer parameters are ideally made in response to the spectral content of the audio The focus on the sonic content of the audio that the user wishes to process is significant The rotary knobs in the traditional mixing desk meets this need and works with a tactile and sensorial relationship with the audio Mycroft and Paterson observes The engineer can place fingers on the dials and adjust them in a way that does not break his or her connection with the sonic artifact but rather takes them into his or her experience of the world 2 The possibility to focus on the audio without having
5. 2 BACHELOR THESIS Percieved Difference in Equalizing Workflow Between a Control Surface and a Mouse and Keyboard What Differences in Workflow do Sound Engineers Perceive Between Rotary Controllers and Mouse and Keyboard when Equalizing with Parametric Ludvig Larsson Bachelor of Arts Audio Engineering Lulea University of Technology Institutionen for konst Kommunikation och larande Perceived difference in equalizing workflow between a control surface and a mouse and keyboard What differences in workflow do sound engineers perceive between rotary controllers and mouse and keyboard when equalizing with parametric equalizer software Ludvig Larsson Lulea University of Technology 2014 Abstract In this research an experiment was made to understand how engineers perceive the differences between using a physical rotary knob and graphical user interface GUI for an equalizing task Subjects were given tasks with control surfaces and with GUIs and a questionnaire to find out if there is a perceived difference in workflow in two different interaction situations with a software parametric equalizer Comparisons based on tasks and questions and overall comparisons are made between the perception of control surface interaction and the mouse and keyboard interaction Different comments from the subjects are discussed and an evaluation of the experiment and influencing variables was made Results show that there are perceived difference
6. ailed critical t value for statistical significance is 2 01 when a 0 05 and df 44 Because three different tests are made on the same set of data a Bonferroni correction had to be made to maintain the statistical significance This means that a is lowered to 0 01 and the two tailed critical t value becomes 2 692 This gives a Statistical difference between the interaction methods in the question about frustration level Quest 2 cS M K CS M K CS M K CS MSE CS IMSE Quest 1 Quest 1 Quest2 Quest2 Quest3 Quest 3 Quest4 Quest4 Quest 5 Quest 5 4 4 2 3 3 3 2 3 3 gt AP ARA AP AA WAR he RB ARA ewe wow ere ee wewwr ere VW wu Ve VW eh Whe Auvaa Ww amp wreuwbtwiwwwwenaebhenrereanawuwuwmuwhtwuuwmwwuw aesh ww S amp S Se SWHeH Se SW DW D NN ce ee ee ee ee RE oe WW KB U NN WW NW wre ww re NN WW WH r Nw AN rr or Ww wre ww wee me we br br NN PURAN wwwe NW me NN Nw We em NW nee ee NN ee ee ee WW ene Kr KN wee Ne WW ee NN WW wwwn EB We UAU NUUA NN BRN NN WH NN NN WW Db ssw AU NNA B amp W swe N NN RR BP RN WW NAIIII NN NN Nn Wn INR WNIYIY Wen NR DN eR rN BN RPh BR NER BR NB WH MN wnre NA WUWU AUN we BB WHNU RR WH HR OM LW OK ROR LW ANN NN Ne SB Wns ner BB ww MN NN LAR LAR OS OS LAR OS RS KAR we es ee AS PAPUNPNPULUN wr we AS hr we LAR ARS PN RR we we wm KS LA NWN We Wwe wen we ee we we we we KR NN LAR ee AR Oe AR ee NN OKI OKI ka Oe Oe Oe Oe Oe Oe PJ m nm bh w ArithmeticMean 3 889 3 756 1 689 2 378 2 511 2 822 1 933 2 400 2 289 2 622 Stand
7. anged To evaluate the perceived workflow several factors of interaction was treated in the questionnaire For example the feeling of certainty understanding how to do what is intended how much time it takes to get to get from start to end in the task and what the end result is or sounds like The number of frequency bands type of filters and equalizer parameters were defined to four bands two with peak filters and two that were switchable between peak and shelving filters Every band has a frequency gain and Q value parameter The definition is based on its simplicity and it is a commonly used design in both classic console EQs and software EQs 5 7 The EQ curve was visual represented when using the control surface and both visual and interactive represented when using the mouse and keyboard Equipment and setup A Behringer BCR 2000 9 was used as control surface in the experiment It contains programmable buttons rotary encoders and the layout was able to adjust to a graphic software parametric EQ with some extra functions The control surface was assigned to a visual representation of a parametric EQ built up in Apple MainStage2 10 as an on screen control See Fig 2 and 3 This was thereafter mapped to a 4 band software parametric equalizer the Waves Renaissance Equalizer 11 in 6 different mixes Each mix had one of the 3 stimuli and one of the 6 tasks ipector Channel 1 Scree Pane
8. ard Deviation 0 682 0 712 0 793 1 051 0 991 0 834 0 837 0 986 0 815 0 806 Tvalue 1232 3 788 592012 Table 1 Table with summed tasks divided by interaction method and questions 10 Control Surface m MM Mouse Keyboard E E E E E E Quest1 Quest2 Quest3 Quest4 Quest5 Fig 6 Graph showing the arithmetic mean of the summed tasks broken down by interaction method and questions The following graphs fig 7 11 are showing comparisons between the control surface and mouse keyboard divided by questions with the arithmetic mean of each task Control Surface Mouse Keyboard TASK 1 TASK 4 TASK 2 TASK 5 TASK 3 TASK 6 Fig 7 Mean of the question How well did you accomplish what you were asked to Control Surface Mouse Keyboard TASK 1 TASK 4 TASK 2 TASK 5 TASK 3 TASK 6 Fig 8 Mean of the question How frustrated were you 11 Control Surface Mouse Keyboard TASK 1 TASK 4 TASK 2 TASK 5 TASK 3 TASK 6 Fig 9 Mean of the question How much time did you have to spend to complete the task Control Surface Mouse Keyboard TASK 1 TASK 4 TASK 2 TASK 5 TASK 3 TASK 6 Fig 10 Mean of the question How insecure were you Control Surface Mouse Keyboard TASK 1 TASK 4 TASK 2 TASK 5 TASK 3 TASK 6 Fig 11 Mean of the question How hard did you have to work to accomplish your level of performance 12 Table 2 is showing the actual time for eac
9. ceived difference in the workflow was unessential but it could also be because of the number of subjects participating and difficulty level of the tasks for example Some other reasons are discussed in the experiment evaluation But in almost all cases except in the bass boost task subjects show a preference for the control surface in the average results in the graphs fig 7 11 Even if it is not statistically significant it is a trend that could give a hint of something and in most cases the subjects were less frustrated insecure and had to work less hard with the control surface than the mouse and keyboard This should not be seen as something to make a clear definition from But it give a good explanation for the answers to the questions asked after all the tasks were finished were there is a significant result and all except one of the 15 subjects prefer the control surface as interaction method When looking at the chart of the average of the actual time it took to complete the tasks and comparing it to what the subjects perceived as time spent to complete the tasks they do have similar trends In the questions asked after all tasks had been completed the mouse and keyboard is clearly seen as more difficult to interact with than the control surface There is also a significant perception that there is a difference in efficiency between the different interaction methods somewhere between a fairly and big efficiency difference due to the scale
10. d bass frequencies with a high pass filter Then try to find unwanted and or resonant frequencies by boosting a band with narrow q value and sweep through the frequencies to find it and then remove a certain amount in that frequency area or areas Thereafter the same procedure could be made to find preferred frequencies and boost them to satisfaction This method is often used to get an EQ setting to work from for an instrument in the mix Method To investigate if there is a perceived difference in the workflow and result between the two different interaction methods of a control surface and mouse and keyboard an experiment consisting of an active user task and a questionnaire was made The active task contained the two different interaction methods for equalizing The subject was given a set of equalizing tasks to complete with each of the interaction methods There were three different tasks each with different stimulus and all the tasks were made with both ways of interaction Both the order of the tasks and the interaction method was randomized With the control surface the subjects could choose to use the screen or not because all parameters was able to perceive from the surface Which give the possibility as previously discussed to focus on the audio without having to watch the screen and at the same time changes the parameters In the key mouse it is mostly necessary to watch the screen to be able to interact and see how parameters are ch
11. d keyboard and Control Surface Mouse Keyboard Fig 13 Showing mean value of Table 3 Table 4 is showing the answers of the question How big difference did you notice in efficiency of your workflow between the two interaction methods 1 Very small 2 Small 3 Fairly 4 big 5 Very big Sujit subj2 4 Subj3 4 Subj Oo SubjS 3 Subj6 4 Ds SubjB o 3 Subj9 5 Sji 3 o Subjtt 3 Subji2 4 o SWB 3 si o SubjtS 3 _ Arithmetic Mean 3 600 Difference in efficiency Standard Deviation 0 632 between the interaction _ HypothecicMean 3 000 methods Twalue 3674 Table 4 Difference in efficiency of workflow Fig 14 Showing mean value of Table 4 14 All except one that is 93 3 of the subjects prefer the interaction of the control surface rather than the mouse and keyboard Which give a statistical significance in a binominal test with a set to 0 05 The cumulative probability is in this case 0 0005 Some comments and motivation around why the questions after all the tasks were finished give a clear statistical significance will be brought up in the discussion section Discussion In the questions answered between each task see Table 1 there is only one question were it is a statistically significant difference between the users perceived interaction and that is the question How frustrated were you This could have several reasons one obvious is that the per
12. f clear experiment instructions was also discovered in the pilot study Subjects The subjects were 15 sound engineer students that were experienced in the use of a parametric equalizer with both a mouse and some kind of pots or rotary controllers Experimental Procedure The instructions were verbally given to the subjects at the same time as all the functions of the control surface the on screen EQ functions etcetera were shown In this test you are going to do a couple of equalizing tasks Then you are going to answer these questions between each task on how you perceived it The task instructions will be given to you on the screen When it says CS you can only use the control surface When it says M K you can only use the mouse and keyboard When using M K everything on the screen can be used but no windows should be moved Everything that is on the CS has a representation on the screen When you feel that you are done with a task you press pause and fill in the questions After the instructions they tried all the functions and the feeling of the different interaction methods out on their own before the actual tasks This to exclude some of the problem with the subject s learning curve in the experiment The audio playback setup was the internal soundcard from a laptop and AKG k240 headphones The rest of the equipment during the experiment was a wireless Logitech mouse and keyboard and the control surface in front of an external
13. h task and subject with results written in seconds There is no statistical significant difference between the interaction methods in the tasks The critical t value is 2 718 df 14 and a 0 016667 with Bonferroni correction The graph fig 12 shows the arithmetic mean of the time in each task DI TASK 1 TASK4 TASK2 TASKS TASK3 TASK6 Subji 42 58 56 54 72 67 Subj2 87 se 29 108 60 121 Subj3 92 216 133 133 137 165 Bobs 153 125 80 67 130 36 bobs 7 160 146 97 1177 123 135 Bosje n 162 149 139 79 76 30 Subj 26 94 23 49 92 104 Subj8 733 1144 57 40 79 88 Subj9 62 145 33 183 96 172 Subj10 87 78 103 55 109 192 Subj11 54 50 109 69 104 108 ubj n 145 149 47 77 27 14 Subji3 51 162 53 26 198 126 Subji4 83 134 187 264 172 158 SubjiS 185 30 39 55 138 111 Mean 97 121 79 87 12 135 St Dev 50 48 58 50 35 Tvalue 1560 069 0255 Table 2 Seconds to complete each task Control Surface Mouse Keyboard TASK 1 TASK 4 TASK 2 TASK 5 TASK 3 TASK 6 Fig 12 Average time to complete each task Results from questions after tasks These were questions asked after all tasks had been completed so it is an overall assessment of the perceived workflow for each subject The critical t value is 2 145 with a 0 05 and df 14 in both Table 3 and 4 which give a statistical significance in both cases 13 Table 3 is showing the answers of the two questions What was your overall impression of difficulty in the interaction with the mouse an
14. heless it has to be considered that the graphics and visual information will affect the workflow and how the users equalize One outcome of this is that it will affect how effective the user does the task that is intended but there are also other aspects The intuitive and creative part of the equalizing workflow also affects the result Because equalizing sometimes is a task guided by subjective preferences the technical visuals graphics and interaction could be a hindrance for the process 2 The purpose with this work is to determine whether the method of interacting with a software parametric equalizer have a significant impact on the perception of the workflow This could give relevant information in how to think around these different interaction systems when equalizing and creating new systems The equalizer interface and interaction In the early days of the digital recording and processing area the concept of control surfaces often was to reduce the physical size of the traditional mixing desk but keep the idea of a similar device used as a remote 3 Today various kinds of control surfaces have been developed for different budgets and tasks generally using the faders and rotary knobs as a basic layout and menu buttons and layers for different functions There is also very common to mix in the box and only use a mouse and keyboard with the computer to adjust all parameters in the mixing process When using mouse and keyboard
15. ho preferred mouse interaction commented that it probably was because of habit and experience Another comment about the advantages with the mouse Keyboard interaction is that it does not have the limitations that come with a control surface Experiment evaluation All subjects were very used to working in the box A more varied subject group may have shown more significant preference differences If the subjects had been some generations older for which the mouse and keyboard not were a natural part of learning the equalizing process it could have been a bigger difference in advantage for the workflow of the control surface The particular plug in and control surface chosen may have impacted the results Though the Waves Renaissance equalizes is widely spread and quite common to use it is not representable for all software equalizers GUI It was chosen because of practical reasons in the experiment but was a little bit oversensitive to mouse movements in the EQ curve as some users perceived This could have affected the perceived workflow for the subjects using this method of equalizing with the mouse and keyboard One comment that confirmed that this might be true was Hate the Waves equalizer specifically but still prefer control surface in general If the EQ curve had been less sensitive to mouse movements may be the differences had been even smaller or more advantages for the mouse and keyboard To be able to change the tasks and
16. ire The questions had the NASA Task Load Index a subjective workload assessment tool 8 as a point of departure With changes and additions to fit this experiment The questions to be answered by the subject between each task was How well did you accomplish what you were asked to How frustrated were you How much time did you have to spend to complete the task How insecure were you How hard did you have to work to accomplish your level of performance Every question had five check boxes with a scale from very bad little to very good much For example How well did you accomplish what you were asked to a a L L L Very bad Badly Acceptable Good Very well See appendix for the whole question forms These questions were showed to the subjects before the experiment so that the result would not be affected between the first and second task After all tasks were completed subjects answered a final set of questions What was your overall impression of difficulty in the interaction with the mouse and keyboard What was your overall impression of difficulty the interaction with the control surface How big difference did you notice in efficiency of your workflow between the two interaction methods These questions also had the scale with five checkboxes Then there were one question of Which interaction method did you prefer and two open questions for the subject to write do
17. l Controls irouped Con lt Controls T Grouped Controls All Controls l o6 fuo Fig 2 Control surface used in experiment Fig 3 Visual eq representation in MainStage The Waves Renaissance Equalizer has the possibility to limit the EQ to four band with a draggable visual EQ curve it can change the filters between bell and shelving and it was possible to hide its unwanted functions below the edge of the screen See Fig 4 lg Subj 15 Tuner Perform Mus Tangentbord G r strang finger ljudet klart och tydligt i elbasen utan a sk ra eller ta bort n gra basfreqvenser i i Lo Shlv Instrumentet High Shiv HM Q RtrnTStart 0 55 00 10 Paused RtrnTStart Copy Pa Gs ee eR Inf In Fig 4 The screen during the experiment Except for the rotary knobs for the four bands there was buttons both on the screen and on the CS for Shelving on off for the highest and lowest frequency band band on off for each frequency band EQ on off play pause jump to start and loop There was also an adjustable volume knob the waveform that is playing and a virtual VU meter In the center there was a textbox explaining the task and which interaction method that was going to be used To get as natural interaction as possible with the Control Surface some parameters were set in main stage The bands were limited to a certain frequency range of Lo Lo mid Hi mid High Because of the limited resolution of 127 step
18. of the questionnaire The perception of how well the subjects accomplish what they were asked to and how much time it took does not show any major differences So the end result and the time it takes to get there may not be especially different all between the interaction methods What you experience on the way to the end result in frustration certainty and work seem to differ a lot more As the two comments from the subjects confirm Both formats have their advantages and disadvantages One can quickly locate frequencies with your mouse while you can do it more carefully with a control surface Can depend on the control surface you re working on but generally I think I prefer the surface over the mouse keyboard 15 It is very task and situation dependent on which method of interaction that is preferable Many times the control surface is not so well adjusted to the actual software equalizer as in this case and in some cases they are indeed One motivation for preferring the control surface was When you did not have to look at the screen it became that I mixed more with the ears than with my eyes Something I very easily does with visual EQ This is something that could be applied to working with analog gear too which may be one of the reasons to preferring the traditional mixing desk equalizer It also confirms the reasoning in 2 of that the technical visuals could be a hinder for the equalizing process The subject w
19. s but they are also task and experience dependent so there is no way of saying that an interaction method is doing the job better than the other What could be concluded are that overall almost all subjects preferred the interaction with the control surface Table of Contents F LT SUL FOL E EIEI EAAS A AATTEITA IOANES PENE EES EIEN EE E EIET OE OET 1 oaao eT En S p NR EEA AAE PA N EEA A E EE E EES N E S ous PA EIEE A A A A 3 The equalizer interface and interaction csssssscsssccsssssesssesssssssesssesssesssessesesesesesesseesesesesanesasereeeseeatess 3 TPES ITA TE CS ceca E ST RR ere EE 4 EO pice satan E PEATA ETE A AS AEA EE EEA E EE TA N A ETN EIEEE TEE AEREE 5 Ed ipmient anid SetU nrpneneprsimis aninga EnEn EREEREER EES EAA ASEEN 5 S e A A E ee ee 6 ON AS E E E E A EE AE A E N E E E AS T ecnesiees 7 CIUE STIO TITT T y N E E E A A E E E E A EE HEN 7 Additional data collected emmsmsssrsssrsrerrrssreerrsrsnr rr rs nr nr nr RR RAR RAR RARE ARR RR RR RR RR RR RR RAR RR RAR RR RAR RR RR RR RR RR RR RR RR RR RAR RR RR RAR RA 8 PLO e Ka ht nearer neers ener renee ener Ae nee een cree eer ee GE eee ee eee ere ett NG 8 STD 1e E ee nn ene A ENNE eee N ATE A A 8 BSS ba Proced G reunnonisanse Ea E eSEE S E IAN NNE Ar RE 8 Results amp AnNa SIS sorreran cee E SEEREN 9 Results from questions between each task cssssssssssesssssssessseesssscssssecscesecssacseacsccaseeesseasseassssesseseeessees 9 Results from Questions after tasks ccs
20. s in the MIDI protocol The Q value and Gain also had a point set as normal were more than one number of MIDI input values represented that point Gain at 0dB and Q value at 0 8 Stimuli Recordings of a grand piano an electric bass and a steel string guitar were used as stimuli for the different tasks The piano was boosted with 13 5 dB at 740 Hz to create at least one obvious resonant frequency The bass was cut with 15 5dB at 3000 Hz to remove the clear finger sounds The acoustic guitar was kept unprocessed All examples were between 10 15 sec so the frequency response did not change too much over time and so the subject could get comfortable with the audio content pretty quick The chosen stimuli were based on the different tasks The audio is before these filters unprocessed recordings that could be found as material to mix in popular music trying to represent a common mix Situation Tasks There was one cutting one boosting and one more subjective and practice related task for each method of interaction This is because variations of the type of task give a more justified picture of the workflow The tasks were Remove the resonant frequency frequencies in the piano Make the string finger sound clear and distinct in the bass guitar without cutting or remove any low end frequencies in the instrument Use the equalizer to improve the sound on this acoustic guitar from your own preferences Questionna
21. screen connected to the laptop with Main Stage running See Fig 5 Fig 5 The experiment situation Results amp Analysis Results from questions between each task Though it is task and dependent if there is a statistical difference between the interaction methods for each question an overall difference is visible if all type of tasks is combined This table Table 1 compares the different interaction methods for each question with summed tasks A number 1 2 3 4 or 5 represent the subject s answer from the checkboxes for each question A paired t test was made between the two interaction methods to see if the difference was Statistically significant The questions in all tables and graphs is marked with Quest 1 How well did you accomplish what you were asked to Quest 2 How frustrated were you Quest 3 How much time did you have to spend to complete the task Quest 4 How insecure were you Quest 5 How hard did you have to work to accomplish your level of performance And the tasks in all tables and graphs is marked with Task 1 and 4 Remove the resonant frequency frequencies in the piano Task 2 and 5 Make the string finger sound clear and distinct in the bass guitar without cutting or remove any low end frequencies in the instrument Task 3 and 6 Use the equalizer to improve the sound on this acoustic guitar from your own preferences Table 1 shows summed tasks divided by interaction method and questions Two t
22. sesssssssssssesessssssssssssscssesecscecscececscaceesseecseseesesseessessessserseessessseees 13 Bet Oo fe eee AIIE I on eee ee Ske Reece One eee Melosh Meee RN SRA SENT BER NEAN he meee 15 Experiment eval alio Drever cement terre rt ine rt eerste otter err ceo RN dr NRA NER 16 F RUNET TOSCA CN P A nso Oc en NR 17 CO TT CIO ES 0 hyena een nner ne eee eee A eee 17 LAA E A bisa E A T PTEE EE EE A ESS ETE EA A NEI EOE E E S E EET 18 Aa R AAEE 19 Introduction Audio engineers using tools with different types of interfaces some physical as knobs and faders which provide tactile feedback We also have purely virtual tools that often require adjustments with a mouse and keyboard In addition our tools function and features the visual and tactile feedback influences how we use it 1 Itis important to understand how visual and tactile information could affect the use of audio tools And since graphical and visual interfaces are becoming more and more important in audio engineering it is vital to understand how this changes the workflow One tool where the visual interface is particularly important is the parametric equalizer This is used in most audio application to for example alter the frequency response of a sound remove resonance or adjust the timbre of instruments In most of the different software parametric EQ designs the underlying signal processing function is basically the same and mostly only the user interfaces are different Nonet
23. take time the researcher was present in the room That could have affected some subject s focus or feeling of time pressure when another person in the room could perceive their actions But the subjects used headphones and was aware of that they were the only ones hearing what they were doing and many of the situations were you ordinarily equalize you are not alone either The answering scale for the questions could have been wider for better resolution to may be able to see other patterns and create proper boxplots and other comparisons Now there could be a difference that did not shown in the graphs and tables because of coarse steps There are however no obvious 16 indicators that data with greater resolution would have led to different conclusions in this study It should be discussed if the EQ curve should be visual and or interactive represented in the graphics of the parametric equalizer It is very common to interact with the help of dragging the EQ curve in the mouse keyboard situation but also common to use the virtual knobs With the rotary controllers you can only interact by turning the rotary knobs and watch an EQ curve With an interactive EQ curve the engineer may have a different workflow This addition would offer more ecological validity but the cost would be control over the variables or the ability to compare interactions in the experiment In this experiment were the result is based on the subject s perception of the in
24. teraction it was chosen to have an interactive EQ curve because of the ecological validity No subjects asked questions about the interaction methods or meaning of the questions after the introduction So there were no visible uncertainties or doubts during the experiment Further research To examine this further other type of equalization situations could be considered were the task is something else than music mix situations Also the equalizing process is something that depends on the audio and other factors around it Could there be a difference in the workflow of equalization with different interaction methods when actually mixing There is also a lot of different situations were a control surface could be used to control plug ins or other parameters in audio production these areas would be interesting to see if there is a different outcome In this research there is many different aspects and considerations to why the control surface is preferred even if the end result does not differ that much it would be interesting to isolate these aspects and examine them further It would also be fascinating to look at the actual difference in equalization results and not just what the user perceive to see if there is any difference in the frequency responses and workflow Conclusions In the summed table which divides the answers by interaction method and not by task there is a statistical significant difference in perceived frustration between
25. the interaction methods to the control surface advantage There are perceived differences but they are also task and experience dependent so there is no way of saying that an interaction method is doing the job better than the other What could be concluded is that almost all subjects preferred one way of interaction overall which is concluded by the questions answered after the tasks This could also have an affect to a certain amount in the long run to enjoy or prefer the tool you work with and not get tired or frustrated after using it during a long period even if the end result and the time it takes to get there do not differ that much 17 References 1 E Bruce Goldstein 1981 The Ecology of J J Gibson s Perception Leonardo Vol 14 No 3 pp 191 195 2 Mycroft J Paterson J 2011 Activity Flow in Music Equalization the Cognitive and Creative Implications of Interface Design AES Convention E Brief 10 Presented at AES 130th Convention London 3 East John W Frindle P A Harrison S I 1995 Philosophy and Implementation of a High End Assignable Control Surface for a Digital Audio Mix System Presented at AES 99th Convention New York Preprint 4123 4 Hlatky M Gohlke K Black D Loviscach J 2009 Enhanced Control of On Screen Faders with a Computer Mouse AES Convention paper 7738 Presented at AES 126th Convention Munich 5 Dewey C Wakefield J 2013 Novel designs for the parametric
26. wn their thoughts If you preferred one of the interaction methods motivate and Other thoughts or comments around the interaction that you want to share All questions and tasks were in the subject s main language Swedish during the experiment to exclude linguistic misunderstandings these are the translations Additional data collected Except for the questionnaire the actual time for each task was captured with a stopwatch The time was captured from the moment the subjects had understood the task instructions and pressed play until they were done with the task This was done for comparisons with the perceived time of interaction and to have one parameter of efficiency to compare the subject s perceived workflow with Pilot study A pilot study was made with three sound engineer students at Lulea University of Technology Each with different main experience with equalizing music studio post production and live This was made to evaluate the experiment and ensure that everything was understood correctly and that nothing was unclear or felt strange This was set up in the same way as the main experiment except for a few improvements that was made after their experiment evaluations The question of the boost task was refined to prevent the subject from cutting in the low frequencies The control surface was given printed frequency Q and gain values for each band before it just had function and colors printed The significance o
Download Pdf Manuals
Related Search