Enhancing speech in reverberation by steady-state suppression

In Proc. of the International Congress on Acoustics, Madrid, 2007

Enhancing speech in reverberation by steady-state suppression

K. Takahashi, K. Yasu, N. Hodoshima, T. Arai and K. Kurisu

Abstract: In order to improve speech intelligibility in reverberation, Arai et al. [Autumn Meet. Acoust. Soc. Jpn., 2001; Acoust. Sci. Tech. 23(8), 2002] proposed steady-state suppression. Our ultimate goal is to implement steady-state suppression in real time by a digital signal processor. For quasi-real-time processing, a consonant enhancement technique in order to compensate for the loudness recruitment was proposed by Yasutake et al. [Tech. Rep. of IEICE Japan, Vol. HIP2005-94, 2005]. These two techniques in Arai et al. and Yasutake et al. are essentially the same in terms of enhancing consonants. Therefore, we simplified the algorithm of steady-state suppression by including the consonant detection method used in Yasutake et al. for quasi-real-time processing. We found the high concordance rate of steady-state portions of both the proposed and the previous methods of steady-state suppression.

[PDF (281 kB)]