摘要
The paper focuses mainly on speech enhancement methods. Because noise may degrade the quality and the intelligibility of speech, various enhancement technologies are applied in speech processing systems. The speech enhancement method in AMR is based on masking properties of the human auditory system, which performs a comb filter before the open loop pitch analysis. During the entire processing course, the filter is employed. While for the unvoiced part, it may degrade speech quality. In view of the energy of the speech, the paper improves this step by using perceptual weighting filter on voiced speech so as to strengthen pitch and harmonic components and suppress the noise because of its periodicity character. The experiment results indicate that the method can improve the quality and intelligibility of the speech while take less computation load.
- 出版日期2009
- 单位华中科技大学