摘要
语音信号处理的速度和结果很大程度上受到端点检测的影响.传统的短时能量和过零率相结合的端点检测算法在低噪声环境下,可以获得满意的检测效果.然而,在高噪声的环境下,甚至难以有效检测语音信号.研究表明,高阶差分结合短时能量的新方法在低噪声环境下,可获得与传统方法相似的效果,而在高噪声下,则明显优于传统方法.
End point detection strongly influences the speed and result of speech signal processing. Traditional methods based on short term energy and zero crossing can obtain satisfactory results under high signal noise ratio conditions, but can not detect speech signal effectively under low noise ratio condition. Research showed that the new method based on high order difference and short term energy can obtain similar results like traditional methods under high signal noise ratio, but outperforms traditional methods under low signal noise ratio.
出处
《北京师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2012年第2期146-148,共3页
Journal of Beijing Normal University(Natural Science)
关键词
端点检测
短时能量
高阶差分
低信噪比
end point detection
short term energy
high order difference
low signal noise ratio
