多通道语音增强算法及应用研究

发布时间：2018-01-23 11:40

本文关键词： 多通道子空间信噪比助听器失真　出处：《福州大学》2014年硕士论文　论文类型：学位论文

【摘要】：语音增强是语音信号处理的一个重要内容,在语音压缩编码和语音识别等领域,有广泛的应用。它是一种噪声抑制技术,对解决语音噪声污染、改进语音通信质量、提高语音可懂度及语音识别率等方面具有重要的意义。随着语音增强技术的发展,传统的单通道语音增强方法,如谱减法、卡尔曼滤波法和子空间法等相继被提出。近年来,由于多麦克风的开发和应用,多通道语音增强方法已倍受关注,已经成为语音增强技术的研究热点。它能在消除噪声、提高语音质量方面优于单通道语音增强方法。同时多通道语音增强技术比单通道较为复杂,这是因为麦克风阵列融合了语音信号的空时信息。多通道语音增强方法研究已得到发展,但一些关键技术有待进一步研究和改进。本文的主要研究工作如下：1.研究现有多通道语音增强技术：如多通道维纳滤波语音增强技术,多通道子空间语音增强技术及基于时空渐进预测的语音增强技术(STP)。通过模拟实现这些多通道语音增强方法,比较在不同种类噪声下的去噪效果,进而分析它们优缺点。2.通过利用多通道子空间语音增强技术在去除噪声上的优势与维纳滤波语音增强技术的平稳特性,本文提出一种基于信号噪声能量比估计的自适应组合维纳滤波器与子空间滤波器的多通道语音增强方法。大量实验表明,在不同的白色噪声及有色噪声下,与传统多通道子空间语音增强技术,如多通道维纳滤波语音增强,基于时空渐进预测的语音增强技术(STP)及信号子空间语音增强技术进行比较,本文提出的自适应联合滤波器的多通道语音增强技术具有较好的语音恢复效果。3.研究了多通道语音增强算法在助听器的应用。本文进一步将提出的多通道语音增强新方法应用于双麦克风助听器系统中,并说明该算法的有效性。
[Abstract]:Speech enhancement is an important part of speech signal processing. It is widely used in speech compression coding and speech recognition. It is a kind of noise suppression technology to solve the problem of speech noise pollution. Improving the quality of speech communication, improving speech intelligibility and speech recognition rate is of great significance. With the development of speech enhancement technology, traditional single-channel speech enhancement methods, such as spectral subtraction. Kalman filtering and subspace methods have been proposed one after another. In recent years, due to the development and application of multi-microphone, multi-channel speech enhancement has been paid more and more attention. It can eliminate noise and improve speech quality, which is better than single-channel speech enhancement method. At the same time, multi-channel speech enhancement technology is more complex than single-channel speech enhancement technology. This is because the microphone array combines the space-time information of the speech signal. The research of multi-channel speech enhancement method has been developed. However, some key technologies need to be further studied and improved. The main research work of this paper is as follows: 1. The existing multi-channel speech enhancement technology, such as multi-channel Wiener filtering speech enhancement technology. Multi-channel subspace speech enhancement technology and spatial-temporal progressive prediction based speech enhancement technology are simulated to achieve these multi-channel speech enhancement methods to compare the denoising effect under different kinds of noise. Then analyze their advantages and disadvantages. 2. Through the use of multi-channel subspace speech enhancement technology in noise removal advantages and Wiener filter speech enhancement technology of the stationary characteristics. In this paper, an adaptive combination of Wiener filter and subspace filter based on the estimation of signal noise energy ratio is proposed for multi-channel speech enhancement. A large number of experiments show that under different white noise and colored noise. Compared with the traditional multi-channel subspace speech enhancement technology, such as multi-channel Wiener filter speech enhancement, space-time progressive prediction based speech enhancement technology and signal subspace speech enhancement technology. The multi-channel speech enhancement technology of adaptive joint filter proposed in this paper has better effect of speech recovery. 3. The application of multi-channel speech enhancement algorithm in hearing aid is studied. The new method of sound enhancement is applied to the dual microphone hearing aid system. The effectiveness of the algorithm is illustrated.
【学位授予单位】：福州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.35

【相似文献】