摘要: 基于计算听觉场景原理,提出了一种混叠语音信号分离算法模型,对两个说话者的混叠声音进行分离。该模型对低频区和高频区的分离分别采用了不同方法,避免了因采用同样方法处理低频高频区而导致对高频段语音不能很好分离的结果。实验结果表明,该模型具有很好的应用效果。
关键词:
计算听觉场景分析,
语音分离,
基音估计,
幅度调制
Abstract: This paper proposes a model of speech separation based on the principle of computational auditory scene analysis, to separate a mixture of two speeches. Different from other models, the model uses different methods to separate resolved and unresolved harmonics, so as to avoid the problem of inability to separate unresolved harmonics successfully. Experimental results show excellent performance of the model.
Key words:
computational auditory scene analysis(CASA),
speech separation,
pitch evaluation,
amplitude modulation
中图分类号:
王 珊;许 刚. 基于计算听觉场景分析的语音混叠信号分离[J]. 计算机工程, 2007, 33(18): 211-213.
WANG Shan; XU Gang. Speech Separation Based on Computational Auditory Scene Analysis[J]. Computer Engineering, 2007, 33(18): 211-213.