Blind Spectral-GMM Estimation for Underdetermined Instantaneous Audio Source Separation

Arberet, Simon; Ozerov, Alexey; Gribonval, Rémi; Bimbot, Frédéric

doi:10.1007/978-3-642-00599-2_94

Cited by 15 publications

(19 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For cSCT, this assumption is replaced by that of at most two predominant sources. Both assumptions do not hold exactly for most audio data [25,26]. Thus, some of the estimated local angular spectra φ(t, f, τ ) do not represent any "true" TDOA possibly leading to poor estimation of the TDOAs.…”

Section: Proposed Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Multi-source TDOA estimation in reverberant audio using angular spectra and clustering

2012

Self Cite

View full text Add to dashboard Cite

-source TDOA estimation in reverberant audio using angular spectra and clustering. Signal Processing, Elsevier, 2012Elsevier, , 92, pp.1950Elsevier, -1960Elsevier, . <10.1016Elsevier, /j.sigpro.2011 Multi-source TDOA estimation in reverberant audio using angular spectra and clustering AbstractWe consider the problem of estimating the time differences of arrival (TDOAs) of multiple sources from a two-channel reverberant audio signal. While several clustering-based or angular spectrum-based methods have been proposed in the literature, only relatively small-scale experimental evaluations restricted to either category of methods have been carried out so far. We design and conduct the first large-scale experimental evaluation of these methods and investigate a two-step procedure combining angular spectra and clustering. In addition, we introduce and evaluate five new TDOA estimation methods inspired from signal-to-noise-ratio (SNR) weighting and probabilistic multi-source modeling techniques that have been successful for anechoic TDOA estimation and audio source separation. The results show that clustering-based methods do not improve upon angular spectrum-based methods. For 5 cm microphone spacing, the best TDOA estimation performance is achieved by one of the proposed SNR-based angular spectrum methods. For larger spacing, a variant of the generalized cross-correlation with phase transform (GCC-PHAT) method performs best.

show abstract

Section: Proposed Methodsmentioning

confidence: 99%

“…Several works [25,26,22,23] have shown that relaxing this assumption can be very beneficial for audio source separation. Thus, we here investigate whether such an approach could be beneficial for multi-source localization.…”

Section: Em Algorithm With Multiple Sources In Each Time-frequency Binmentioning

confidence: 99%

Multi-source TDOA estimation in reverberant audio using angular spectra and clustering

2012

Self Cite

View full text Add to dashboard Cite

show abstract

“…Note that this example is a difficult, real-world mixture, which involves several sources mixed in the center (bass, singing voice, certain drums) and several harmonic sources with comparable pitch range (singing voice, 11 The bass is modeled as a sum of 4 sources to facilitate initialization, since we do not know a priori its spatial direction. The drums are modeled as a sum of 4 sources for the same reason, but also because the drum track is often composed of several sources (e.g., snare, hi-hat, cymbals, etc) that can be mixed in different directions.…”

Section: ) State-of-the-artmentioning

confidence: 99%

A General Flexible Framework for the Handling of Prior Information in Audio Source Separation

Ozerov

Vincent

Bimbot

2012

IEEE Trans. Audio Speech Lang. Process.

Self Cite

211

336

View full text Add to dashboard Cite

Abstract-Most of audio source separation methods are developed for a particular scenario characterized by the number of sources and channels and the characteristics of the sources and the mixing process. In this paper we introduce a general audio source separation framework based on a library of structured source models that enable the incorporation of prior knowledge about each source via user-specifiable constraints. While this framework generalizes several existing audio source separation methods, it also allows to imagine and implement new efficient methods that were not yet reported in the literature. We first introduce the framework by describing the model structure and constraints, explaining its generality, and summarizing its algorithmic implementation using a generalized expectation-maximization algorithm. Finally, we illustrate the above-mentioned capabilities of the framework by applying it in several new and existing configurations to different source separation problems. We have released a software tool named Flexible Audio Source Separation Toolbox (FASST) implementing a baseline version of the framework in Matlab.

show abstract

“…Finally, in section 6, we evaluate the performance of our approach on mixtures of monophonic, polyphonic and percussive music sources and compare it to the state-of-the-art approaches. Preliminary aspects of this work were presented in [28] in the case of Spectral-GMM source model and linear instantaneous mixtures. In this paper we extend it to convolutive mixtures and to other Spectral models (Spectral-GSMM and Spectral-NMF), and we provide a more consistent experimental evaluation.…”

Section: Organization Of the Papermentioning

confidence: 99%

A tractable framework for estimating and combining spectral source models for audio source separation

et al. 2012

Self Cite

View full text Add to dashboard Cite

International audienceThe underdetermined blind audio source separation (BSS) problem is often addressed in the time-frequency (TF) domain assuming that each TF point is modeled as an independent random variable with sparse distribution. On the other hand, methods based on structured spectral model, such as the Spectral Gaussian Scaled Mixture Models (Spectral-GSMMs) or Spectral Non-negative Matrix Factorization models, perform better because they exploit the statistical diversity of audio source spectrograms, thus allowing to go beyond the simple sparsity assumption. However, in the case of discrete state-based models, such as Spectral-GSMMs, learning the models from the mixture can be computationally very expensive. One of the main problems is that using a classical Expectation-Maximization procedure often leads to an exponential complexity with respect to the number of sources. In this paper, we propose a framework with a linear complexity to learn spectral source models (including discrete state-based models) from noisy source estimates. Moreover, this framework allows combining di erent probabilistic models that can be seen as a sort of probabilistic fusion. We illustrate that methods based on this framework can significantly improve the BSS performance compared to the state-of-the-art approaches

show abstract

Blind Spectral-GMM Estimation for Underdetermined Instantaneous Audio Source Separation

Cited by 15 publications

References 14 publications

Multi-source TDOA estimation in reverberant audio using angular spectra and clustering

Multi-source TDOA estimation in reverberant audio using angular spectra and clustering

A General Flexible Framework for the Handling of Prior Information in Audio Source Separation

A tractable framework for estimating and combining spectral source models for audio source separation

Contact Info

Product

Resources

About