一种基于听觉掩蔽效应的语音增强方法

为提高增强语音的听觉效果 ,研究了一种基于听觉掩蔽效应的语音增强方法。推出了一个功率谱域的基于听觉掩蔽效应的不等式准则 ,并用这个准则动态地选择一个作为语音短时谱幅度估计器的非线性函数的参数值 ,通过这个参数自适应变化的非线性函数对语音谱幅度进行估计实现语音增强。在此基础上 ,设计实现了一个单声道语音增强算法。对增强语音的客观测试和非正式听音测试表明 :相对于传统的减谱法和对数短时谱幅度最小均方误差估计增强法 ,基于听觉掩蔽效应的语音增强方法能更好地抑制背景噪声

Speech enhancement method based on auditory masking

This paper presents a speech enhancement method based on the masking properties of the human auditory system and a non linear filter. The speech phase is ignored to identify enhanced criterion in the power spectral domain based on masking. The criterion is then applied to determine the parameter values for a non linear filter employed to estimate the short time spectral amplitude of speech. The method is then used to develop a single channel speech enhancement algorithm. Objective measurements combined with informal subjective listening tests show that the proposed algorithm can suppress audible noise more effectively than the popular power spectral subtraction method and the minimum mean square error log spectral amplitude estimation method.