1 / 27

The voices in your head & utilizing head movement in hearing-aid signal processing

The voices in your head & utilizing head movement in hearing-aid signal processing. Alan Boyd ( CeSIP /IHR) Supervisors Prof. John Soraghan ( CeSIP ) & Wm. Whitmer (IHR). 21 st February 2013. Internalization/externalization. Initial aims and objectives.

hedva
Download Presentation

The voices in your head & utilizing head movement in hearing-aid signal processing

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The voices in your head&utilizing head movement in hearing-aid signal processing Alan Boyd (CeSIP/IHR) Supervisors Prof. John Soraghan (CeSIP) & Wm. Whitmer (IHR) 21st February 2013

  2. Internalization/externalization

  3. Initial aims and objectives • find the causes of internalization in hearing-impaired listeners • study the effect of hearing-aid signal processing on internalization • determine the prevalence of the perception of internalization in the hearing-impaired population • improve hearing-aid signal processing to reduce internalization

  4. Externalization I • Externalized condition • Internalized condition 30° ← 3 m → Render headphones ‘acoustically transparent’ No HRTF but still ITD 30° ← 3 m →

  5. Results for 1 talker Boyd et. al, JASA (2012)

  6. Results for 4 talkers Boyd et. al, JASA (2012)

  7. Externalization II Impulsive noise bursts 30° 2 m Listener with/without hearing-aids 2 m Continuous single male talker

  8. Externalization II • Normal hearing • Hearing impaired

  9. Internalization survey • Participants • 267 respondents: 70 without hearing aids; 122 unilaterally aided; 75 bilaterally aided. • Prevalence • Percentage of respondents who experience internalization: 19.6% unilaterally-aided • 15.7% • without hearing aids 35.1% bilaterally-aided

  10. Psychoacoustic research summary • Hearing-impaired have a compressed perception of externalization • Results suggest that internalization occurs when auditory cues are not prominent, with no effect of the hearing aid. • Survey results show an increase in the prominence of internalization with increasing number of hearing aids • No scope for a signal processing solution...

  11. f1-attentuation TOTAL FAILURE

  12. Head movement & hearing aids • Listeners move • The position of the ears (and hearing aids) relative to a sound source can change due to head movements • Hearing aids work best when static • Performance of adaptive noise-reduction algorithms can be reduced by head movements or moving sources • Head movements may provide useful information • Changing user behaviour may require a different setting • Algorithms could compensate for head movements

  13. Hardware • MEMS triple-axis gyroscope (ITG 3200) • Provides angular velocity information • Programmed via Arduino and USB serial • Small enough to fit on a hearing aid • Patents exist for basic use of a gyroscope in hearing aids gyroscope Arduino

  14. Version 1 • Program selection based on user behaviour • Directional programs are useful for listening to one sound source in noisy environments • Omni-directional programs are useful for localizing sounds • Gyroscope information is used to smoothly mix between programs: • Directional when listener is stationary • Omni-directional when listener is “searching”

  15. Adaptive directional mic array Teutsch & Elko, 2001

  16. Behavioural switching A A A A B B B B A DIR GYRO OMNI GYRO DIR B

  17. Version 3 • Compensates for head-movements during direction of arrival (DOA) estimates • Uses generalized cross-correlation (GCC-PHAT) Binaurally-linked mics DOA histogram bins SOURCE1 SOURCE2

  18. Version 3 - System Mic 1 Mic 2 Gyroscope STFT STFT Find angle rotated since previous time-step Cross-correlation Over-sampled IFFT Shift current histogram against rotated angle Find correlation peak delay → angle of arrival Update histogram

  19. Version 3 – One source Boyd et. al, Proc. ICA (2013)

  20. Version 3 – Four sources in reverb Boyd et. al, Proc. ICA (2013)

  21. Version 3 – One source in noise Boyd et. al, Proc. ICA (2013)

  22. Version 4 Position 1 • Front/back detector • GCC-PHAT measures ± 90° • Measurement in front and rear • hemisperes is the same • Short measurement period • Compares peak movement to head movement • ↓ source angle + clockwise rotation = Position 1 • ↑ source angle + clockwise rotation = Position 2 Cone of Confusion Position 2

  23. Improvements & combinations • Improve gyroscope performance using accelerometers & magnetometers (Razor Attitude and Heading Reference System) • Combine gyro system with NavinChatlani’ssteerable beamformer • Recordinteraural time differences, level differences and head movements in real-world situations (ISRA 2013)

  24. Conclusion • Perception of internalization of sound occurs in a significant minority of the hearing-impaired population • The hearing aid may not cause it, however it may exacerbate the perception if it already exists • Head movement information can be used for; • Automatic program selection based on behaviour • Robust DOA estimates • Longer timescale techniques

  25. Thanks for listeningAny questions?

  26. Externalization over headphones

  27. Head movement & hearing aids A A A A A B B B B B Gyroscopically guided beamformer Standard adaptive beamformer DIR GYRO OMNI GYRO DIR

More Related