论文部分内容阅读
文本相关重点研究的说话人语音识别是语音识别研究中的一个重要分支。本论文介绍了一种采用特征空间轨迹匹配方式的基于数字语音的身份识别系统的设计方法。并针对“0”到“9”这十个数字,对不同说话人的语音特征的可区分性进行了探讨,提出了数字区分度算法,以便找出最能有效区分不同说话人的数字,进而在识别系统中提高识别效率。实验结果表明,本论文提出的数字区分度算法在小规模语音说话人集合中得到了验证。
Speech related speech recognition is an important branch of speech recognition research. This dissertation introduces a design method of digital speech recognition system based on feature space matching. And for the “0 ” to “9 ” these ten numbers, the speech characteristics of different speakers were discussed, the number of discrimination algorithm proposed in order to find the most effective distinction between different speakers The number of recognition system, and then improve the recognition efficiency. Experimental results show that the algorithm proposed in this dissertation is validated in small speech speaker sets.