论文部分内容阅读
话者鉴别的一个基本前提是“个体间的语音差异大于个体的语音变异”。通过对30对双胞胎正常朗读语音的共振峰特性、韵律特性、长时LPC谱特性进行定量测量、比较、分析,统计得出了这些特性在检验双胞胎语音中的置信度、判阈及其区别力。
One of the basic prerequisites for speaker identification is that “the voice difference between individuals is greater than the individual’s voice variation.” Through the quantitative analysis of the formative, prosodic and long-term LPC spectra of 30 pairs of normal reading speech, the confidence, the judgment threshold and the discrimination of these characteristics in the test of twin speech were compared, analyzed and statistically analyzed .