Alpha-stable low-rank plus residual decomposition for speech enhancement

Simsekli, Umut and Erdoğan, Halil and Leglaive, Simon and Liutkus, Antoine and Badeau, Roland and Richard, Gael (2018) Alpha-stable low-rank plus residual decomposition for speech enhancement. In: IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Calgary, AB, Canada

Full text not available from this repository. (Request a copy)


In this study, we propose a novel probabilistic model for separating clean speech signals from noisy mixtures by decomposing the mixture spectra into a structured speech part and a more flexible residual part. The main novelty in our model is that it uses a family of heavy-tailed distributions, so called the α-stable distributions, for modeling the residual signal. We develop an expectation-maximization algorithm for parameter estimation and a Monte Carlo scheme for posterior estimation of the clean speech. Our experiments show that the proposed method outperforms relevant factorization-based algorithms by a significant margin.
Item Type: Papers in Conference Proceedings
Uncontrolled Keywords: Alpha-stable distributions; Audio source separation; Monte Carlo Expectation-Maximization; Speech enhancement
Divisions: Faculty of Engineering and Natural Sciences
Depositing User: Halil Erdoğan
Date Deposited: 27 May 2023 22:57
Last Modified: 27 May 2023 22:57

Actions (login required)

View Item
View Item