We revisit the source image estimation problem from blind source separation (BSS). We generalize the traditional minimum distortion principle to maximum likelihood estimation with a model for the residual spectrograms. Because residual spectrograms typically contain other sources, we propose to use a mixed-norm model that lets us finely tune sparsity in time and frequency. We propose to carry out the minimization of the mixed-norm via majorization-minimization optimization, leading to an iteratively reweighted least-squares algorithm. The algorithm balances well efficiency and ease of implementation. We assess the performance of the proposed method as applied to two well-known determined BSS and one joint BSS-dereverberation algorithms. We find out that it is possible to tune the parameters to improve separation by up to 2 dB, with no increase in distortion, and at little computational cost. The method thus provides a cheap and easy way to boost the performance of blind source separation.
DOI: 10.21437/Interspeech.2020-2158
Cite as: Scheibler, R. (2020) Generalized Minimal Distortion Principle for Blind Source Separation. Proc. Interspeech 2020, 3326-3330, DOI: 10.21437/Interspeech.2020-2158.
@inproceedings{Scheibler2020, author={Robin Scheibler}, title={{Generalized Minimal Distortion Principle for Blind Source Separation}}, year=2020, booktitle={Proc. Interspeech 2020}, pages={3326--3330}, doi={10.21437/Interspeech.2020-2158}, url={http://dx.doi.org/10.21437/Interspeech.2020-2158} }