Low complexity Bayesian single channel source separation

Abstract

We propose a simple Bayesian model for performing single channel speech separation using factorized source priors in a sliding window linearly transformed domain. Using a one dimensional mixture of Gaussians to model each band source leads to fast tractable inference for the source signals. Simulations with separation of a male and female speaker using priors trained on the same speakers show comparable performance with the blind separation approach of Jang and Lee [1] with a SNR improvement of 4.9 dB for both the male and female speaker. Mixing coefficients can be estimated quite precisely using ML-II, but the estimation is quite sensitive to the accuracy of the priors as opposed to the source separation quality for known mixing coefficients which is quite insensitive to the accuracy of the priors. Finally, we discuss how to improve our approach while keeping the complexity low using machine learning and CASA approaches [1,2,3,4].

Adaptation of Bayesian Models for Single-Channel Source Separation and its Application to Voice/Music Separation in Popular Songs

“…Better separation performance may be achieved with the MAP criterion (12) and appropriate priors compared to what can be obtained with general models. Fig.…”

“…To illustrate this issue, let us take an example where one of the sources is a voice signal (as, for instance in [5], [7], [10], and [12]). Either the voice model has been trained on a particular voice but its generalization ability tends to be poor to other voices, or it is trained on a group of voices but then it requires a large number of parameters, and even though, it tends to lack selectivity to a particular voice in a particular mix, not to mention the variability problems that can be caused by different recording and/or transmission conditions.…”

“…We suppose that the parameters of model are independent from those of model and that the pdf of the parameters of each model depends on the parameters of the corresponding general model, which can be summarized as . Finally, we have the following MAP criterion: (12) Note that the MAP criterion (12), in contrast to the ML criterion (11), involves the prior pdfs , which forces the adapted models to stay attached to the general ones. Thus, the general models play the role of a priori knowledge about the sources.…”

“…This expression reflects two following ideas. 1) Model adaptation corresponds to the attachment of the adapted models to the general models, for instance by means of prior pdfs , as in (12). 2) The use of missing acoustic data corresponds to the fact that the model parameters are estimated from the mix , whereas the actual acoustic data (the sources ) are unknown (i.e., missing).…”

Principles and Typical Computational Limitations of Sparse Speaker Separation Based on Deterministic Speech Features

“…Such conditions are that there be at least as many input channels (microphones) as sound sources, that background noise be essentially absent, and that there be an instantaneous mixing of the sources (i.e., no time delays). Even sophisticated variants like the noisy or convolutive ICA (Hyvärinen et al, 2001) are bound by rather restrictive conditions, although single-channel ICA methods have been reported to be able to separate a male from a female speaker (Jang & Lee, 2003;Beierholm, Pedersen, & Winther, 2004). ICA's underlying assumption is that the signal is a linear mixture of the contributing sources that are independent realizations of a random process.…”