Cauchy Multichannel Speech Enhancement with a Deep Speech Prior
Cauchy Multichannel Speech Enhancement with a Deep Speech Prior
复制标题
具有深度语音先验的柯西多通道语音增强
DOI：
10.23919/eusipco.2019.8903091
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
A. Liutkus
中科院分区：
文献类型：
--
作者：
Mathieu Fontaine;Aditya Arie Nugraha;R. Badeau;Kazuyoshi Yoshii;A. Liutkus
我们提出了一种基于概率模型的半监督多通道语音增强系统，该模型假设语音和噪声都服从重尾多变量复柯西分布。正如我们所提倡的，这允许处理强烈和不利的噪音条件。因此，该模型是由源幅度谱图和源空间散射矩阵参数化。为了处理散射矩阵的非可加性，我们的第一个贡献是在投影空间上执行增强。然后，我们的第二个贡献是结合联合收割机的潜变量模型的语音，这是通过以下的变分自动编码器框架，与低秩模型的噪声源进行训练。在测试时，应用迭代推理算法，该算法产生用于分离的估计参数。首先从噪声语音中估计语音潜变量，然后通过梯度下降法进行更新，同时使用一种优化均衡策略来更新两个源的噪声和空间参数。我们的实验结果表明，柯西模型优于国家的最先进的方法。标准差分数也表明，该方法对非平稳噪声更鲁棒。
We propose a semi-supervised multichannel speech enhancement system based on a probabilistic model which assumes that both speech and noise follow the heavy-tailed multi-variate complex Cauchy distribution. As we advocate, this allows handling strong and adverse noisy conditions. Consequently, the model is parameterized by the source magnitude spectrograms and the source spatial scatter matrices. To deal with the non-additivity of scatter matrices, our first contribution is to perform the enhancement on a projected space. Then, our second contribution is to combine a latent variable model for speech, which is trained by following the variational autoencoder framework, with a low-rank model for the noise source. At test time, an iterative inference algorithm is applied, which produces estimated parameters to use for separation. The speech latent variables are estimated first from the noisy speech and then updated by a gradient descent method, while a majoriation-equalization strategy is used to update both the noise and the spatial parameters of both sources. Our experimental results show that the Cauchy model outperforms the state-of-art methods. The standard deviation scores also reveal that the proposed method is more robust against non-stationary noise.
影响因子：
4.3
作者：
Heymann, Jahn;Drude, Lukas;Haeb-Umbach, Reinhold
通讯作者：
Haeb-Umbach, Reinhold
影响因子：
4.3
作者：
Vincent, Emmanuel;Watanabe, Shinji;Marxer, Ricard
通讯作者：
Marxer, Ricard