摘要
在语音识别中,噪声严重影响语音特征提取,使得正确率明显下降。针对这一情况,提出了子带加权平均语音识别算法(Sub-Band Weighted Average Speech Recognition,SBWASR),在噪声环境下,该方法能有效地提高语音识别的准确性。最后通过实验验证特定条件下该方法是正确有效的。
The voice feature extraction is affected badly by noise, and the ratio of speech recognition is dropped obviously. According to this situation, Sub-band weighted average Speech Recognition （SBWASR） is propounded in this paper. It can effectively improve the veracity of speech recognition in noisy environment. It is also validated through experiments that the method is correct and effective in specific conditions.
出处
《三明学院学报》
2009年第4期386-390,共5页
Journal of Sanming University
基金
福建省自然科学基金(2009J01296)
福建省大学生创新性实验项目(ZL0708/CS)
关键词
语音识别
特征提取
噪声
子频带
speech recognition
feature extraction
noise
sub-band
