噪声环境下说话人识别的组合特征提取方法

针对在干净语音环境下识别率很高的说话人识别系统,在噪声环境下识别率显著降低的缺点,本文结合具有多分辨率分析特点的小波变换技术,提出一种基于小波变换的组合特征提取算法,以提高说话人识别系统在噪声环境下的识别性能。对40个说话人的语音库SUDA2002-D2,在噪声环境下进行的识别实验结果表明,本文提出的组合特征提取算法可以在噪声环境下有效地提高说话人识别系统的识别性能。

A Combined Feature Extraction Method for Speaker Identification under Noisy Conditions

A speaker recognition system with high performance in relatively clean environment will become deficient with unac- ceptable recognition performance in noisy environment.Wavelet transform holds multi-resolution analysis abilities.In this paper,a new combined-feature-extraction algorithm based on wavelet transform is proposed to improve the recognition rate of speaker identification in noisy conditions.Experiments on SUDA2002-D2 Chinese speech corpus show that the proposed algorithm is quite efficient for speaker i- dentification in noisy conditions.