Abstracts Submitted: 510
Number of Users: 710
View Abstracts Submitted
Back to home Page
It often becomes an issue that wind noise annoys hearing aid users in outdoor situations since it masks speech signals and speech intelligibility is degraded. Wind noise is suppressed to attach a windscreen when a general microphone is used. Meanwhile, attaching it to hearing aids is difficult owing to cosmetic reasons. A wind noise cancellation system using the characteristics of head-related transfer functions had been proposed by authors to resolve these problems. It can effectively suppress wind noise, preserving binaural cues. Here we call the system “FFT-based system”. However, it requires 32 ms frame processing (meaning 512 points frame processing) so as to maintain high resolution in frequency. On the other hand, it is known that the tolerable group delay for mild hearing loss should be below about 5 ms. This issue had been resolved by using frequency warping filter-bank with the analysis length of 32 points as synthesis and analysis stages of our system. The frequency resolution of low-frequency part was kept an efficient resolution by adjusting warping parameter in order to achieve that of FFT-based system with 512 points frame length. In this paper, it is described that the FWF-based low-delay binaural wind noise cancellation system is implemented on a prototype hearing assistive device to evaluate real-time performance of the system. This system applies that the number of all-pass filters for frequency warping is set to 31, and also frame length of FFT analysis is 32. For objective evaluations, the spectrogram of speech and wind noise in non-processed and processed signals, and SNR improvements between I/O signals are calculated. From these results, it is clarified that this system can suppress wind noise even on the actual device and the output SNRs is improved compared with input SNRs.
© Copyright 2017 All Rights Reserved