摘要
在分析各个Mel子带抗噪性能的基础上,提出了鲁棒性的说话人识别算法,经实验结果证明,在噪声环境下此算法能有效地提高说话人识别系统的识别率。
The text-independent speaker recognition is an important part of speaker recognition now, it plays a more and more important role because of its ease-to-use and highly popular application in the information technology. At present, the accuracy of recognition is very high in clean environment, but the performance in noisy environment needs to be improved. Based on the analysis of the noise robustness of each subband, a robust recognition algorithm was presented and the simulation shows the algorithm could increase the accuracy of recognition in noise environment.
出处
《苏州大学学报(工科版)》
CAS
2007年第4期4-7,共4页
Journal of Soochow University Engineering Science Edition (Bimonthly)
基金
国家自然科学基金资助项目(编号60572076)
江苏省高校自然科学基金资助项目(编号05KJB510113)
关键词
说话人识别
与文本无关
矢量量化
Mel子带
speaker recognition
text-independent
vector quantization
Mel-Subband