基于改进VQ算法的文本无关的说话人识别

基于矢量量化的说话人识别,因其运算过程简单等特点,在文本无关的说话人识别领域有着广泛的应用。论文根据说话人识别中训练语音的特点并结合快速搜索算法,对矢量量化的码书形成算法进行了改进,提出了一种基于改进算法的与文本无关的说话人识别方法。经实验结果证明,论文的方法加快了码书的形成,减少了码书形成的计算量,改善了码本的性能,提高了说话人识别的识别率。

Text-independent Speaker Recognition Based on Improved VQ Algorithm

Vector quantization(VQ),without regard to the difficult speech segment and the complex time warping,is a useful and simple method for text-independent speaker recognition.Considering the character of the speaker training data,this paper proposes a method to improve the algorithm of codebooks forming,and puts forward a method of text-independent speaker recognition based on the improved algorithm.The simulation shows the algorithm can greatly reduce the time of forming codebooks,improve the performance of the codebooks and increase the correct rate of speaker recognition.