As single channel speech enhancement, common methods based on short-time spectral amplitude estimation such as spectral-subtraction and Wiener filtering are considered to be simple and effective for suppressing the additive noise in noisy environments. However, these techniques may distort the high frequency harmonics of speech, which are important to the language understanding.

To solve this problem, under the guidance of Academician HOU Chaohuan, CAI Yu and YUAN Jianping of Institute of Acoustics, Chinese Academy of Sciences conducted a series of studies and proposed a new approach to deal with the above question. In their proposal, they first use a comb filter in the time domain based on Wiener filtering to enhance the sonant part of the language and restrain the residual noise left after the Wiener filtering. Then they implement another comb filter in frequency domain to retrieve the high frequency harmonics that have been compressed. By these two steps, speech signal is clearer and more understandable. The practical experiment result shows that this new method is an improvement for harmonic enhancement and noise reduction.

The research result was published on the recently issued journal of Chinese Journal of Scientific Instrument (Volume 31, Issue 1, January 2010, Pages 26-31).