摘要
为了评价不同环境下语言清晰度的恶化程度,提出了一种基于包络回归的语言清晰度客观评价方法。使用普通话语音作为测试信号,使用Gammatone滤波器组对语音进行听觉预处理。利用包络回归模型计算包络的变化特性,从而客观地评价普通话语音清晰度。将实验结果与传统测试方法STIPA作对比,该方法比STIPA有更好的拟合度。
In order to evaluate the deterioration of speech intelligibility in different environments,an objective evaluation method of speech intelligibility based on envelope regression is proposed.The Mandarin speech is used as a test signal,and the speech is pre-processed using the Gammatone filter bank.The envelope regression model is used to calculate the variation characteristics of the envelope,so as to objectively evaluate the speech intelligibility of Mandarin.The experimental results are compared with the traditional test method STIPA,which has a better degree of fitting than STIPA.
作者
张海峰
刘璜
ZHANG Haifeng;LIU Huang(School of Electronics Information,Hangzhou Dianzi University,Hangzhou 310018,China)
出处
《传感器与微系统》
CSCD
2020年第1期50-52,共3页
Transducer and Microsystem Technologies