语音失真测度系统的构成与算法实现
Composition and Arithmetic Realization of Speech Distortion Measurement System
-
摘要: 介绍了语音失真测度系统的构成以及实现的算法思想.采用前端处理技术,实现了对重放语音信号进行分帧、端点检测等特征的提取,获得了以Mel尺度倒谱参数作为衡量语音失真测度的特征向量.采用动态时间弯折算法,获得了用于语音质量客观评价的语音失真测度.Abstract: Composition and arithmetic realization of speech distortion measurement system are introduced in this paper.Mel-scaled Frequency Cepstrum Coefficients are used as the characteristic vectors by some characteristics withdrawal for the replay speech signal,such as cutting frames and checking the initial point etc.The system adopts the algorithm of dynamic time warping to acquire the speech distortion used for the objective evaluation of the speech quality.