Switching Variational Auto-Encoders for Noise-Agnostic Audio-visual Speech Enhancement - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Switching Variational Auto-Encoders for Noise-Agnostic Audio-visual Speech Enhancement

Résumé

Recently, audiovisual speech enhancement has been tackled in the unsupervised settings based on variational autoencoders (VAEs), where during training only clean data is used to train a generative model for speech, which at test time is combined with a noise model, e.g. nonnegative matrix factorization (NMF), whose parameters are learned without supervision. Consequently, the proposed model is agnostic to the noise type. When visual data are clean, audiovisual VAE-based architectures usually outperform the audio-only counterpart. The opposite happens when the visual data are corrupted by clutter, e.g. the speaker not facing the camera. In this paper, we propose to find the optimal combination of these two architectures through time. More precisely, we introduce the use of a latent sequential variable with Markovian dependencies to switch between different VAE architectures through time in an unsupervised manner: leading to switching variational auto-encoder (SwVAE). We propose a variational factorization to approximate the computationally intractable posterior distribution. We also derive the corresponding variational expectation-maximization algorithm to estimate the parameters of the model and enhance the speech signal. Our experiments demonstrate the promising performance of SwVAE.
Fichier principal
Vignette du fichier
robust_vae.pdf (229.86 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03155445 , version 1 (01-03-2021)

Identifiants

Citer

Mostafa Sadeghi, Xavier Alameda-Pineda. Switching Variational Auto-Encoders for Noise-Agnostic Audio-visual Speech Enhancement. ICASSP 2021 - 46th International Conference on Acoustics, Speech, and Signal Processing, Jun 2021, Toronto / Virtual, Canada. pp.1-5, ⟨10.1109/ICASSP39728.2021.9414097⟩. ⟨hal-03155445⟩
124 Consultations
315 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More