语音信号中相位信息的听觉感知研究

通过主观听觉测试实验，研究了语音信号中相位信息对人的听觉感知的影响．实验结果表明，保持语音信号的幅度谱不变，在改变其相位谱时，只要重建信号在时域中的包络不变，重建语音和原始语音就不存在主观听觉上的差异．重建语音的听觉感知效果主要取决于附加相位对频率的导数的起伏幅度．重建语音中不同频率分量之间的最大相对时移决定语音感知的质量，当最大相对时移小于10ms时，语音感知质量最优；只要相位失真带来的不同频率分量之间的最大相对时移小于20ms，就不会影响对连续语音的正常理解．

Study of Auditory Perception of Phase Information in Speech Signal

The perceptual effect of the phase information in speech has been studied by auditory subjective tests. It has been shown that when the phase spectrum in speech is changed and it's amplitude spectrum is kept constant, there is no different auditory perception between the original speech and the reconstructed speech if the envelope of the reconstructed speech signal is not changed. The auditory perception effect of the reconstructed speech is mainly dependent on the amplitude of the derivative of the additive phase with respect to frequency. The perceptual quality in speech is determined by the maximum relative time shift between different frequency components. When the maximum relative time shift is less than 10 ms, the perceptual quality is excellent. The speech understanding will not be depressed provided that the maximum relative time shift between different frequency components due to phase distortion is less than 20 ms.