基于修正Mel域掩蔽模型和无语音概率的耳语音增强

Speech enhancement based on modified Mel masking model and speech absence probability in whispers

提出了一种基于修正Mel域听觉掩蔽模型和无语音概率的耳语音增强方法。该方法根据耳语音的发音特点对Mel频率进行修正,对每一帧耳语音信号进行Mel域频带滤波,同时通过无语音概率(SAP)动态地确定每个频带的听觉掩蔽阈值,对不同的听觉掩蔽阈值自适应地调整谱减系数来进行耳语音增强。对增强后的耳语音进行客观和主观测试,结果表明,该方法与其它谱减法相比,能将残留噪声和背景噪声控制在人耳掩蔽阈值下,取得更小的语音失真,主观听觉也得到了很大的改善。

A method of whispered speech enhancement using auditory masking model in modified Mel-domain and Speech Absence Probability （SAP） is proposed. In light of the phonation characteristic of whispered speech, we modify the Mel Frequency Scaling model. Whispered speech is filtered by the proposed model. Meanwhile, the value of masking threshold for each frequency band is dynamically determined by speech absence probability. Then whisper speech enhancement is conducted by adaptively rectifying the spectrum subtraction coefficients using different masking threshold values. Results of objective and subjective tests on the enhanced whispered speech signal show that compared with other methods, the proposed method can enhance whispered speech signal with better subjective auditory quality and less distortion by reducing the music noise and background noise under the masking threshold value.