声音转换技术在语音识别预处理技术中的应用
首发时间:2009-04-30
摘要:本文提出将声音转换技术应用于语音识别预处理技术中,通过修改待识别语音的共振峰参数、基音频率参数和平局幅度参数,使待识别语音从音色、音调、音量三个方面减小与语音库中标准模板语音间的声学特征差异。ABX测试、基于LPCC参数测试和基于MFCC参数测试的结果表明:改进的预处理技术能有效地减小待识别语音与标准模板语音间的声学特征差异。
For information in English, please click here
Voice Conversion in Voice Recognition Pretreatment
Abstract:In this paper, a new improved voice recognition pretreatment method based on voice conversion is presented. By modifying the formant, pitch frequency and average amplitude, the acoustic differences between identify voice and standard voice will be decreased. ABX evaluation, LPCC measurement and MFCC measurement indicate that the improved pretreatment method can decrease the acoustic differences effectively.
Keywords: formant pitch frequency average amplitude voice conversion pretreatment
基金:
论文图表:
引用
No.3186846542412410****
同行评议
共计0人参与
勘误表
声音转换技术在语音识别预处理技术中的应用
评论
全部评论0/1000