摘要说话人识别在自动身份鉴别方面具有重要的现实意义。文章在分析了说话人识别实现的可行性基础上*提出采用长时线性预测倒谱 +,--.,)系数和矢量量化 /0)模型的不依赖于文本的说话人识别方法，并成功地地应用于一个数据库查询系统中-Abstract Speaker recognition has important practical significance in terms of automatic authentication. The article analyzes achieve the feasibility of speaker recognition based on* the linear prediction using long inverted spectrum+-.) Coefficient and vector quantization/0) model does not rely on the text of the speaker recognition, and successfully applied to a database query system