Sontacchi A., Noisternig M., Majdak P., Höldrich R.

Transcription

1 $Q2EMHFWLYHRGHORI/RFDOLVDWLRQLQ %LQDXUDO6RXQG5HSURGXFWLRQ6\VWHPV Sontacchi A., Noisternig M., Majdak P., Höldrich R. $(6VW,QWHUQDWLRQDO&RQIHUHQFH -XQH 6W3HWHUVEXUJ5XVVLD

2 ,QVWLWXWHRI(OHFWURQLFXVLF DQG$FRXVWLFV University of Music and Dramatic Arts Graz, Austria KWWSLHPNXJDFDW LQFROODERUDWLRQZLWK $.*$FRXVWLFV9LHQQD$XVWULD KWWSZZZDNJFRP

3 Outline HRIR based Binaural Sound Reproduction Systems using Ambisonic Definitions Mathematical Model Results Outlook

4 Binaural Systems using Ambisonic Binaural Reproduction Systems: filtering virtual source signals with HRIRs incorporate head tracking to improve source localisation Problem: time-varying interpolation between HRIRs

5 Binaural Systems using Ambisonic Why Ambisonic? head movement: - time-invariant HRIR filter - cheap time-variant rotation matrix decoding is independent from coding number of transmit channels is independent from number of virtual sources

6 Binaural Systems using Ambisonic encode signals depending on source position rotate Ambisonic depending on head position decode Ambisonic to virtual speaker signals convolve speaker signals with HRIRs

7 Definitions localisation function localisation error localisation blur average localisation error average localisation blur localisation blur localisation error

8 Mathematical Model Intentions: classifying Binaural Reproduction Systems prediction of localisation performance Main assumption: the reference HRIRs result in optimal localisation 2D systems only

9 Mathematical Model Parameter Mathematical Model Device under test ILD ITD HRIRs ILD ITD ITDA ILDA Result Merger

10 Interaural Time Difference (ITD) 2 2 Amplitudengang Magnitude Response.3 HRIR, ϕ= HRIR for Amplitude in db x -3 x Gruppenlaufzeit Group Delay Time Amplitude Time in s x -4 FFT Phase in rad Frequenz in Hz Frequency in Hz x x 4 Phasengang Phase Response π Gruppenlaufzeit Group Delay Time in in s s GΦ GI Frequenz in Hz Frequency in Hz x 4 x Frequenz in Hz Frequency in Hz x 4 x 4

11 Interaural Time Difference (ITD) Calculation of the group delay time: impulse response (HRIR) zero phase filter over critical bands energy center of gravity per band group delay time per band Group Delay Time [s] x Group Delay Time -4 Frequency [Hz] x 4

12 Interaural Time Difference (ITD) ITD over critical bands: difference of the group delay time between ipsi- and contralateral eardrum Group Delay Time left ear right ear ITD critical bands azimuth angle

13 Interaural Level Difference (ILD) Calculation of the ILD: amplitude response (HRTF) energy-difference over critical bands between ipsi- and contralateral eardrums energy over critical bands left ear right ear ILD critical bands azimuth angle

14 Interaural Difference Angle (ITDA/ILDA) Search Algorithm: 5 x -4 ƒ,7'irurqhfulwlfdoedqg FDOFXODWHGIURPPHDVXUHG +5,5 GLVWRUWHG,7' UHIHUHQFH,7' ITD [s] Deviation Azimuth Angle [ ]

15 Merging of ITDA and ILDA ITDA: fade out above 8 Hz ILDA: fully weighted Gewichtung der ITD.8 ITDA Gewichtungsfaktor Frequenzgruppe in bark Gewichtung der ILD.8 Result ILDA Gewichtungsfaktor Frequenzgruppe in bark

16 Results () localisation function (averaged over critical bands) localisation blur (standard deviation over critical bands) Θ Θ + Θ Θ = Θ = = ), ( ) ( ) ( 2 ), ( ) ( ) ( 2 ) ( ],/',/',/' ],7',7',7' ] ] Z ] Z ] ] Z ] Z / [ ] = = Θ Θ Θ = Θ 24 2 ) ( ), ( ) ( ) ( 2 ) ( ] L L,7',/' L L / ] ] Z ] Z %O

17 Results (2) average localisation blur (averaged over the azimuth angle) / = [ /( ΘL ) ΘL ] L= 2 average localisation error averaged over azimuth angle referring to the MAA %O = 36 %O( Θ) Θ= $$ ( Θ)

18 Analysis Variations of design parameters: reference HRIRs length of HRIR ambisonic order weighting of ambisonic channels number of virtual speaker

19 HRIR Type () azimuth angle [ ] 2 azimuth angle [ ] Kemar-HRIR (#2) Propr.-HRIR (#9) filter length = 28 samples 5-5 localisation error [ ] localisation blur [ ] Kemar-HRIR (#2) Proprietary-HRIR (#9) filter length = 28 samples

20 HRIR Type (2) 6 ORFDOLVDWLRQHUURU 6.5 ORFDOLVDWLRQEOXU localisation error [ ] rd ord. KEMAR, 5 (#) 3rd ord. KEMAR, 5 (#2) 3rd ord. AKG, 5 (#9) 4th ord. KEMAR, 5 (#8) 4th ord. AKG, 5 (#23) localisation blur [multiples of MAA] rd ord. KEMAR, 5 3rd ord. KEMAR, 5 3rd ord. Propr., 5 4th ord. KEMAR, 5 4th ord. Propr., filter length [samples] filter length [samples]

21 Ambisonic Order () azimuth angle [ ] azimuth angle [ ] th ord. (#8) 3rd ord. (#) 2nd ord. (#) - localisation error [ ] localisation blur [ ] 4th ord. (#8) 3rd ord. (#) 2nd ord. (#) filter length = 28 samples

22 Ambisonic Order (2) ORFDOLVDWLRQHUURU ORFDOLVDWLRQEOXU th ord. (#8) 3rd ord. (#) 2nd ord. (#) th ord. (#8) 3rd ord. (#) 2nd ord. (#) localisation error [ ] localisation blur [multiples MAA] filter length [samples] filter length [samples]

23 Ambisonic Weighting () azimuth angle [ ] 5 azimuth angle [ ] 3 6 3rd ord., Kaiser window (#24) 4th ord., fully weighted (#8) 4th. ord. Kaiser window (#22) filter length = 28 samples 5-5 localisation error [ ] localisation blur [ ] Kaiser window, 3rd order (#24) Fully weighted, 4th order (#8) Kaiser window, 4th order (#22) length of filter = 28 samples

24 Ambisonic Weighting (2) ORFDOLVDWLRQHUURU ORFDOLVDWLRQEOXU localisation error [ ] rd ord., weighting:.4 (#3) 3rd ord., fully weighted (#) 3rd ord., Kaiser window (#24) 4th ord., fully weighted (#8) 4th ord., Kaiser window (#22) localisation blur [multiples of MAA] rd ord., weighting:.4 (#3) 3rd ord., fully weighted (#) 3rd ord., Kaiser window (#24) 4th ord., fully weighted (#8) 4th ord., Kaiser window (#22) filter length [samples] length of filter [samples]

25 Conclusion prediction of localisation is possible results correspond with ambisonic theory - ambisonic not less than 3rd order for correct localisation - weighting of channels using Kaiser-window yields better localisation - filter length up to 28 samples model was evaluated using listening tests

26 Outlook incorporate IGD (LQWHUDXUDOJURXSGHOD\) as a cue for high frequency localisation classification regarding coloration and externity extend the model for 3D applications warping tables to predict and minimise the localisation errors for further implementations

27 ,QVWLWXWHRI(OHFWURQLFXVLF DQG$FRXVWLFV 7KDQN\RX University of Music and Dramatic Arts Graz, Austria KWWSLHPNXJDFDW