摘要： 噪音是降低语音识别系统精度的关键因素，因此，如何从带噪语音信号中提取出有效的语音特征是提高语音识别系统识别率的重要途径。该文在分析语音特征提取方法的基础上提出改进算法。实验表明，采用LDA+MLLT+CMS算法组合提取出的语音特征具有较好的鲁棒性，在噪音环境下的平均音节识别率为43.79%。该组合在中文大词汇量连续语音识别系统中也有较好的性能，音节识别率达到83.56%。

关键词: 特征提取, 主分量分析(PCA), 线性区分分析(LDA), 语音识别

Abstract: Noise is a pivotal factor that reduces recognition rate of a speech recognition system. So how to extract effective speech characteristics becomes an important path for a speech recognition system to increase accuracy. This paper analyses speech feature extraction and makes improvement of it. Experimental results indicate that the algorithm combined with LDA+MLLT+CMS has better robustness than other combinations. Average syllable recognition rate reaches 43.79% by using it under conditions of noises. The algorithm combination has also a good performance in Mandarin Large Vocabulary Continuous Speech Recognition (LVCSR). Syllable recognition accuracy achieves 83.56%.

Key words: feature extraction, Principal Component Analysis(PCA), Linear Discriminant Analysis(LDA), speech recognition

中图分类号: 

• TP311