A study on signal extraction in noisy and reverberant environment

本研究在人类心理声学和听觉生理学知识的基础上，探讨了语音增强和语音分离的模型。利用消去模型对语音进行增强。特别注意使用空间滤波技术降低噪声，并使用频率滤波技术提高基频估计的鲁棒性。这两种技术都采用了消去模型的概念。此外，利用Bregman提出的启发式规则约束，克服了分离两个声源的问题。仿真结果表明，空间滤波和频率滤波都能有效增强语音。因此，这些滤波方法可以有效地用于自动语音识别系统的前端，以及语音特征提取。声音分离模型可以精确地从噪声信号中提取出期望的信号，即使是在波形中。此外，本研究还探讨了基于哺乳动物听觉生理数据的声源方向估计模型。该模型可以解释神经放电传递时间和相位信息与耳际时差准确性之间的关系。

This research discusses models of speech enhancement and segregation based on knowledge about human psychoacoustics and auditory physiology. The cancellation model is used for enhancing speech. Special attention is paid to reducing noise by using a spatial filtering technique, and increasing the robustness of fundamental frequency estimation by using a frequency filtering technique. Both techniques adopt concepts of the cancellation model. In addition, some constraints related to the heuristic regularities proposed by Bregman are used to overcome the problem associated with segregating two acoustic sources. Simulated results show that both spatial and frequency filtering are useful in enhancing speech. As a result, these filtering methods can be used effectively at the front-end of automatic speech recognition systems, and for speech feature extraction. The sound segregation model can precisely extract a desired signal from a noisy signal even in waveforms.Additionally, this research discusses models of sound source direction estimation based on physiological data of mammal audition. The model can explain the relationship between transmission of temporal and phase information by nerve firing and accuracy of interaural time differences.