一种基于线性判别分析的稳健话音检测方法

A Robust Speech/Non-speech Detection Based on LDA

给出一种新的话音检测方法,即在SNR算法的基础上,应用线性判别分析(LDA)对语音特征参数进行降维。在大噪声环境下,该方法提高了系统的稳健性。同时将这种新的方法与基于信噪比(SNR)和基于噪声/语音统计量(N&SSTAT)的算法做了比较,实验表明该方法可以提高检测效率。

In speech recognition, a speech / non-speech detection must be robust to noise. In this work, a new method for speech / non-speech detection using a Linear Discriminant Analysis (LDA) which bases on SNR algorithm is presented. Applying LDA can depress the dimension of the speech feature parameter. In the noisy environment, this method improves the robust of system. This new algorithm is compared to the algorithms based on signal to noise ratio (SNR) and noise/speech statistic (N&S STAT).