摘要
基于语音信号的频谱特性,本文对说话人识别技术中Mel倒谱参数做了改进,并通过Microsoft Visual C++6.0验证了在低信噪比时使用改进后的Mel倒谱参数可以提高说话人识别系统的正确识别率。
This article improves the algorithm of Mel-frequency cepstral coefficients in speaker recognition technology based on frequency character of speech signal. The improved Mel-frequency cepstral coefficients are proved to raise the recognition accuracy of speaker recognition system in low SNR condition by experiments with Microsoft Visual C ＋＋ 6.0.
出处
《计算机与现代化》
2007年第11期26-28,共3页
Computer and Modernization
关键词
说话人识别
Mel倒谱参数
speaker recognition
Mel-frequency cepstral coefficients
