由于说话人的语音特征和个性特征到目前为止无法很好地分离,本文提出了基于语音分类的说话人多维特征的提取方法,将语音识别技术应用到说话人特征提取上,提取出的N维组合特征较其它普通特征有更高的有效性。该方法从汉语语音的特点出发,对基于汉语的说话人识别进行研究。实验结果表明它的有效性较长时平均特征的有效性提高了2.915%。 Since the speaker’s voice features and personality traits have not been well separated so far, this paper proposes a speaker-based multidimensional feature extraction method based on speech classification. The speech recognition technology is applied to the speaker feature extraction and the extracted N-dimensional Combined features are more effective than other common features. Based on the characteristics of Chinese speech, this method studies the speaker recognition based on Chinese. The experimental results show that the validity of the average feature is improved by 2.915% when its validity is longer.