Abstract: The performance of speech quality evaluation for speech coding has been very clear, but it may not be applied to face mask speech. The suitability of speech quality evaluation measures under various noise environments in the application of air speech and face mask speech is discussed. Mean opinion score and three kinds of objective speech quality evaluation measures are taken to evaluate noisy speech and enhanced speech with various signal-to-noise, which include segmental signal-to-noise ratio, modified bark spectral distortion and perceptual evaluation of speech quality, and the suitability is judged by the accordance with subjective evaluation. Wiener filtering and LSA-MMSE algorithms are taken to enhance the noisy speech. Pink noise and wave noise are used. The simulation results show that the suitability of speech quality evaluation algorithms is limited to the kind of speech, SNR of noisy speech, background noise environment and the kind of speech enhancement algorithm. With regard to pink noise, PESQ is not suitable for evaluating air speech enhanced by wiener filtering, and MBSD is only suitable for evaluating face mask speech enhanced by LSA-MMSE. Under the environment of wave noise, PESQ can evaluate face mask speech, and MBSD is not suitable for evaluating face mask speech.

Key words: face mask speech, Wiener filtering, Modified Bark Spectral Distortion（MBSD）, Perceptual Evaluation of Speech Quality（PESQ）

摘要： 针对语音编码的音质评价算法性能已十分明确，但对于面罩语音不一定适用。讨论了语音质量评价算法对空气语音与面罩语音在不同噪声环境下的适用性。采用主观意见得分和三种客观评价测度对多种信噪比的带噪语音和增强语音进行评价，包括分段信噪比、改进的巴克谱失真（MBSD）和语音感知质量评价（PESQ），根据与主观评价的一致性判断客观评价方法的适用性。增强算法采用维纳滤波法和对数谱最小均方误差法（LSA-MMSE），噪声采用粉红噪声、海浪噪声。仿真结果表明，语音质量评价算法的适用性与语音类型、信噪比、背景噪声、增强算法种类有关。粉红噪声环境下，PESQ不适合评价经维纳滤波增强的空气语音；MBSD算法只适用于评价经LSA-MMSE增强的面罩语音。海浪噪声环境下，PESQ适用于评价面罩语音，MBSD不适合评价面罩语音。

关键词: 面罩语音, 维纳滤波, 改进巴克谱失真（MBSD）, 语音感知质量评价（PESQ）