基于仿生模式识别的语音识别研究

The Speech Signal Recognition Based on Bionic Pattern Recognition

与传统模式识别的方法相比,仿生模式识别能识别单模板,并大大减小误识率。基于这两个优点,将仿生模式识别应用于语音信号处理中,进行2次识别。第1次使用传统模式识别方法,第2次为单模板识别。对小词汇量的关键词识别实验结果表明能够大大降低误识率。

Compared with the traditional pattern recognition, bionic pattern recognition can recognize single template and reduce error rate. Based on the two merits, the bionic pattern recognition is applied to speech signal processing with the use of secondary identification. Firstly,it uses traditional methods of pattern recognition, secondly, it uses single template recognition. Small vocabulary keyword spotting experimental results show a significantly lower error rate.