摘要
端点检测是语音信号处理过程中非常重要的一步,它的准确性直接影响到语音信号处理的速度和结果。提出了基于ICA(独立分量分析)增强和谱熵的语音端点检测方法。仿真实验表明此方法快速有效,具有较强的抗噪能力,特别适合低信噪比下的端点检测。
The endpoint detection which influences the speed and results of speech signals processing directly is a important step for speech signal processing. In this paper, a new speech endpoint detection method based on ICA （Independent Component Analysis） enhancement and spectral entropy is proposed. The simulating results show that the proposed method can detect the endpoint exactly in the low SNR environments.
出处
《电声技术》
2006年第10期49-50,54,共3页
Audio Engineering
关键词
独立分量分析
谱熵
端点检测
independent component analysis
spectral entropy
endpoint detection
