【信息技术】【2011.12】基于最小统计谱减法的单通道语音增强

【信息技术】【2011.12】基于最小统计谱减法的单通道语音增强_第1张图片
本文为瑞典布莱津理工大学(作者:Md. Zameari Islam)的硕士论文,共59页。

言语是人类交往的基本源泉。在通信过程中,语音信号的质量和可懂度通常会因周围的噪声而降低。因此,损坏的语音信号需要增强,以提高质量和清晰度。在语音处理领域,人们致力于发展语音增强技术,通过减少干扰噪声来恢复语音信号。

本文主要研究了一种基于最小统计量的单通道语音增强技术,该技术通过谱减法进行降噪。最小统计量是指通过求出噪声信号的平滑功率谱的最小值来估计非平稳噪声信号的功率谱,从而避开了语音活动的检测问题。谱减法的性能评估是使用单通道语音数据和各种噪声水平的噪声类型。该评价方法是为了寻找最优的参数值,从而改进该算法,使其更适合于语音通信目的。在MATLAB中实现了该系统,并通过考虑不同的性能指标、不同的信噪比改善(SNRI)和频谱失真(SD)对系统进行了验证。对于不同的滤波器组设置,例如不同的子带数目以及不同的抽取和插值比率,计算SNRI和SD。该方法在SNRI和SD性能方面提供了有效的语音增强。

Speech is an elementary source of humaninteraction. The quality and intelligibility of speech signals duringcommunication are generally degraded by the surrounding noise. Corrupted speechsignals need therefore to be enhanced to improve quality and intelligibility.In the field of speech processing, much effort has been devoted to developspeech enhancement techniques in order to restore the speech signal by reducingthe amount of disturbing noise. This thesis focuses on a single channel speechenhancement technique that performs noise reduction by spectral subtractionbased on minimum statistics. Minimum statistics means that the power spectrumof the non-stationary noise signal is estimated by finding the minimum valuesof a smoothed power spectrum of the noisy speech signal and, thus, circumventsthe speech activity detection problem. The performance of the spectralsubtraction method is evaluated using single channel speech data and for a widerange of noise types with various noise levels. This evaluation is used inorder to find optimum method parameter values, thereby improving this algorithmto make it more appropriate for speech communication purposes. The system isimplemented in MATLAB and validated by considering different performancemeasure and for different Signal to Noise Ratio Improvement (SNRI) and SpectralDistortion (SD). The SNRI and SD were calculated for different filter banksettings such as different number of subbands and for different decimation andinterpolation ratios. The method provides efficient speech enhancement in termsof SNRI and SD performance measures.

  1. 引言
  2. 项目背景与相关工作
  3. 基于最小统计的谱减法
  4. 具体实现与结果
  5. 结论

更多精彩文章请关注公众号:【信息技术】【2011.12】基于最小统计谱减法的单通道语音增强_第2张图片

你可能感兴趣的:(【信息技术】【2011.12】基于最小统计谱减法的单通道语音增强)