一种鲁棒特征提取算法的设计和实现

Design and Implement of Robust Feature Extraction Algorithm

提出了一种用于语音识别的鲁棒特征提取算法,这种算法基于最小方差无失真响应(MVDR)谱估计技术,它在Mel频率尺度上估计MVDR谱,并对得到的MVDR谱进行调制谱滤波,然后提取其倒谱系数作为特征参数。使用这种算法设计了一个抗噪孤立词语音识别系统,在汽车噪声,人群噪声和高斯白噪声三种噪声环境下,与传统算法按多种信噪比做了对比实验。实验结果表明该系统在这三种噪声环境下的识别率均得到了不同程度的提高。

A robust feature extraction algorithm for speech recognition was proposed, This algorithm is based on the Minimum Variance Distortionless Response （MVDR） spectrum estimation method. It estimates MVDR spectrum at Mel frequency scale and filters the modulation spectrum of the MVDR spectrum, then the cepstrum coefficients are extracted as the feature parameter. A robust isolated word speech recognition system based on this algorithm was designed. Experiments were conducted to compare the proposed algorithm with traditional feature extraction algorithms under different levels of car noise, babble noise and gauss white noise, The results indicate that the recognition accuracy of this system has been improved at some degrees under the three noise conditions.