Speech enhancement is an important field of speech processing. It is an effectivemethod to solve noise pollution and improve speech quality. So, it is important to pursuean effective way for evaluation of speech enhancement algorithms.We introduce the current state of development, application and research of speechenhancement algorithms and objective assessment of speech quality. Since SNR can notreflect the characteristic of speech variety with time, we explore in this paper thedistance measure for evaluation of speech enhancement algorithm performance. Weselect several speech enhancement algorithms and compare their performances bycomparing the distance measure we choose (such as SegSNR, BSD and Mel-CD). We dothis though three aspects such as continuous speech, different part of continuous speechand isolated phonemes. The results of evaluation for speech enhancement algorithmsprove the method we use is effective. |