Binaural speech segregation system on single board computer
Treść / Zawartość
A pocket-size binaural speech segregation system has been developed and assembled with available consumer hardware. It can enhance a target speech in a certain direction while attenuating interfering sounds from other directions. This system is based on the frequency domain binaural model (FDBM) and it segregates multiple speeches based on the directivity. This real-time system is implemented on a low-cost single board computer which might be used as a hearing assistance device. Performance of the system is evaluated in a normal laboratory room as well as anechoic chamber. Even in a room with reverberation, the system works well and show the almost same performance obtained in anechoic chamber.
Bibliogr. 6 poz., fot. kolor., rys., wykr.
- 1. H. Nakashima, Y. Chisaki, Y. Usagawa, M. Ebata, Frequency domain binaural model based on interaural phase and level differences, Acoustical Science and Technology, 24 (2003) 172 - 174.
- 2. Y. Chisaki, K. Matsuo, K. Hagiwara, H. Nakashima, T. Usagawa, Real-time processing using the frequency domain binaural model, Applied Acoustics, 68 (2007) 923 - 938.
- 3. I. W. Selesnick, Short-time fourier transform and its inverse, 2009. (In: url http://eeweb.poly.edu/iselesni/EL713/STFT/stft_inverse.pdf visited on 06/24/2019).
- 4. N. Hiruma, R. Kouyama, H. Nakashima, Y. Fujisaka, Low delay wind noise cancellation for binaural hearing aids. INTER-NOISE, (2016) 4844 - 4854.
- 5. MIT-Media-Lab, HRTF Measurements of a KEMAR dummy-head microphone, 2019. (https://sound.media.mit.edu/resources/KEMAR.html visited on 06/24/2019).
- 6. J. Agnew, J. M. Thornton, Just noticeable and objectionable group delays in digital hearing aids, J. the American Academy of Audiology, 11 (2000) 330 - 336.