摘要
针对英文语音发音标准化评价准确性不高的问题,提出一种基于发音特征倒谱系数感知的英文语音发音标准化的模式识别对比方法。首先构建英文语音发音的语音信号采集模型,对采集的英文语音信号进行发音器官的动作属性配对描述。然后提取英文语音发音信号的倒谱特性,采用梅尔频率倒谱系数感知方法进行英文语音发音特征建模和发音位置及方式的模式识别,为语音发音提供标准化对比模式。最后进行实验分析,测试结果表明,采用该方法进行英文语音发音特征检测和模式识别的准确度较高,对发音特征的声学建模有效可靠。
Aiming at the problem that the evaluation accuracy of English pronunciation standardization is not high,a pattern recognition and contrast method for English pronunciation standardization based on the perceptual standard of pronunciation characteristic cepstral coefficient is proposed in this paper.Speech signal acquisition model of English voice pronunciation is built to pair and describe action attributes of pronunciation organ for the acquired English voice signal first,and then extract the cepstrum characteristics of English speech signal.The sensing method of Mel frequency cepstral coefficients is used to carry out English speech pronunciation feature modeling,and recognize the pattern of pronunciation position and mode,so as to provide a standardized comparison mode for voice pronunciation.The experimental analysis and the test results show that the proposed method is effective and reliable for the acoustic modeling of the pronunciation features,and has the high accuracy for the feature detection and pattern recognition of English speech pronunciation.
作者
郑碧君
刘涛
ZHENG Bijun;LIU Tao(Wuchang Shouyi University,Wuhan 430064,China;Central China Normal University,Wuhan 430079,China)
出处
《现代电子技术》
北大核心
2017年第12期28-30,共3页
Modern Electronics Technique
基金
国家自然科学基金(11101170)
关键词
英文发音
语音信号
模式识别
发音标准化评价
English pronunciation
speech signal
pattern recognition
pronunciation standardization evaluation