摘要
提高汉语浊音基频实时提取精度是语音识别的关键技术之一。本文论述了汉语浊音基频实时提取的意义、难度和误差分析,提出—种“过零检测,信号分离”的新方法。它能有效地减小工频调制误差,振幅起落误差和背景噪声误差,从而达到提高基频量测精度之目的。
Increasing the accuracy in real-time extraction of fundamental pitch from Chinese voiced sounds is one of key techniques in speech recognition.This paper deals with the significance,difficulty and er- ror analysis of real-time extraction of fundamental pitch from Chinese voiced sounds.and,it puts for- ward a″Zero-crossing detection,signal separation″method,This method can decrease effectively the er- rors arising from power frequency modulation,amplitude fluctuation and background noise in order to improve the accuracy in measuring the fundamental pitch.
出处
《浙江师大学报（自然科学版）》
1993年第1期85-89,共5页
Journal of Zhejiang Normal University(Natoral Sciences)
关键词
基频
实时提取
汉语
元音识别
Chinese speech recognition
real-time extraction of fundamental pitch
Zero-crossing detection
threshold voltage
amplitude fluctuation error
signal separation.
