Deep neural network based multichannel audio source separation

Aditya Arie Nugraha 1 Antoine Liutkus 2 Emmanuel Vincent 1
1 MULTISPEECH - Speech Modeling for Facilitating Oral-Based Communication
Inria Nancy - Grand Est, LORIA - NLPKD - Department of Natural Language Processing & Knowledge Discovery
2 ZENITH - Scientific Data Management
LIRMM - Laboratoire d'Informatique de Robotique et de Microélectronique de Montpellier, CRISAM - Inria Sophia Antipolis - Méditerranée
Abstract : This chapter presents a multichannel audio source separation framework where deep neural networks (DNNs) are used to model the source spectra and combined with the classical multichannel Gaussian model to exploit the spatial information. The parameters are estimated in an iterative expectation-maximization (EM) fashion and used to derive a multichannel Wiener filter. Different design choices and their impact on the performance are discussed. They include the cost functions for DNN training, the number of parameter updates, the use of multiple DNNs, and the use of weighted parameter updates. Finally, we present its application to a speech enhancement task and a music separation task. The experimental results show the benefit of the multichannel DNN-based approach over a single-channel DNN-based approach and the multichannel nonnegative matrix factorization based iterative EM framework.
Type de document :
Chapitre d'ouvrage
Audio Source Separation, Springer, A Paraître
Liste complète des métadonnées

Littérature citée [77 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01633858
Contributeur : Aditya Arie Nugraha <>
Soumis le : lundi 13 novembre 2017 - 14:30:56
Dernière modification le : vendredi 12 janvier 2018 - 01:53:27

Fichier

nugraha_book18.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01633858, version 1

Citation

Aditya Arie Nugraha, Antoine Liutkus, Emmanuel Vincent. Deep neural network based multichannel audio source separation. Audio Source Separation, Springer, A Paraître. 〈hal-01633858〉

Partager

Métriques

Consultations de la notice

131

Téléchargements de fichiers

109