计算机工程与应用
計算機工程與應用
계산궤공정여응용
COMPUTER ENGINEERING AND APPLICATIONS
2009年
31期
128-129,133
,共3页
基频检测%小波去噪%归一化AMDF
基頻檢測%小波去譟%歸一化AMDF
기빈검측%소파거조%귀일화AMDF
pitch detection%wavelet-denoising%normalized AMDF
提出一种对舍噪语音进行基频检测的新方法.先对舍噪语音进行小波去噪,然后再经过预处理后,采用归一化的AMDF算法对语音进行基频提取,后期对基频信号采用搜索试探方法进行平滑处理,通过实验表明,该方法比传统方法有更好的鲁棒性,尤其在低信噪比的情况下.
提齣一種對捨譟語音進行基頻檢測的新方法.先對捨譟語音進行小波去譟,然後再經過預處理後,採用歸一化的AMDF算法對語音進行基頻提取,後期對基頻信號採用搜索試探方法進行平滑處理,通過實驗錶明,該方法比傳統方法有更好的魯棒性,尤其在低信譟比的情況下.
제출일충대사조어음진행기빈검측적신방법.선대사조어음진행소파거조,연후재경과예처리후,채용귀일화적AMDF산법대어음진행기빈제취,후기대기빈신호채용수색시탐방법진행평활처리,통과실험표명,해방법비전통방법유경호적로봉성,우기재저신조비적정황하.
This paper proposes a pitch detection method for the noisy speech.Firstly,the wavelet transform is used to de-noising the noisy speech,then after the pre-processing,the normalized AMDF is adopted to extract the speech's fundamental frequencies. During the post-treating,the searching method is used to smooth the fundamental frequencies.Experimental results show the new method has better robust than the traditional method,especially in the low snr.
