Abstract: An improved spectral subtraction algorithm based on codebook learning for speech enhancement in non-stationary noise conditions is proposed. The proposed algorithm contains two steps. The priori information about the spectrum of speech and noise is modeled using autoregressive model and the speech and noise codebooks are constructed. The speech and noise are estimated in each time frame by solving a log-spectral distortion minimization problem. The proposed algorithm can adapt to varying levels of noise even while speech is present. On the other hand, autoregressive modeling results in smooth frequency spectrums and thus reduces musical noise. Experimental results show that the proposed algorithm outperforms the traditional spectral subtraction algorithm and multiband spectral subtraction algorithm.

Key words: speech enhancement, spectral subtraction, noise estimation, auto regressive model

摘要： 提出一种可适应非平稳噪声环境的基于码本学习的改进谱减语音增强算法。该算法分为训练阶段和增强阶段。训练阶段，使用自回归模型对语音和噪声的频谱形状进行建模并构造语音和噪声码本；增强阶段，采用对数谱最小化算法估计出语音和噪声的频谱，通过谱相减消除噪声。算法在每个时间帧估计语音和噪声频谱，即使在语音存在时仍能够有效跟踪快速变化的非平稳噪声；采用自回归模型能得到噪声频谱的平滑估计，减少了音乐噪声。实验仿真表明，相比于传统谱减法和多带谱减法，改进的谱减法具有更好的噪声抑制性能并且语音失真更小。

关键词: 语音增强, 谱减法, 噪声估计, 自回归模型