Abstract：It is obtained that every formant can be traced and found out by studying and analyzing the Chinese speech in the way of experiment in this paper, combining with the generally articulation feature of the Chinese speech, a speech enhancement algorithm based on tracing the formant is proposed. The speech frames can be determined from the non-speech frames effectively using the algorithm, and then, all the noise in the non-speech frames can be wiped off simply and validly, and noise in the speech frame can be degraded largely. Also, the algorithm has some other good characteristics, such as simple computation and noise-environment transplantability and so on.
2009, Vol. 31 
