双谱图在语音分析中的应用

Application of bispectrum in speech analysis

为便于对不同的语音特征进行辨析,提出了一种基于双谱分析的语音辨别方法。通过采样语音数据,进行相应处理,再进行双谱分析,对不同类型的语音双谱图进行对比,因其语音包含因素不同,双谱图形上表现的双谱特征亦不同,其结果差异明显。试验结果表明:将双谱应用于语音分析中,可以对不同语音进行识别,能提高语音辨别的准确性,对处理语音信号有显著效果。双谱图是进行语音辨识的一种有效方法,具有良好的应用前景。

In order to analyze on phonetic characteristics of different speaker,put forward a method for speech discrimination based on bispectrum analysis. Sampling data of speech,corresponding processing,and carry out bispectrum analysis,comparing bispectrum of different speech,because of speech that contains different factors,characteristics of bispectrum is also different,the difference of result is obvious. Experimental results show that the apply bispectrum in speech analysis,can realize speech recognition of different speaker. It can greatly improve the accuracy of speech recognition,and it has a significant effect on processing of speech singles. Bispectrum is an effective method for speech recognition,and it has good application prospect.