Audio-visual speech enhancement with a deep Kalman filter generative model

TL;DR

This paper introduces an audiovisual deep Kalman filter model for speech enhancement that effectively leverages the sequential nature of speech and visual data, outperforming existing VAE-based models in noisy conditions.

Contribution

The paper proposes a novel AV-DKF generative model that incorporates sequential modeling for audiovisual speech enhancement, improving performance over non-sequential models.

Findings

AV-DKF outperforms audio-only models in noisy environments.

Sequential modeling enhances speech reconstruction quality.

The proposed method effectively fuses visual and audio data for better results.

Abstract

Deep latent variable generative models based on variational autoencoder (VAE) have shown promising performance for audiovisual speech enhancement (AVSE). The underlying idea is to learn a VAEbased audiovisual prior distribution for clean speech data, and then combine it with a statistical noise model to recover a speech signal from a noisy audio recording and video (lip images) of the target speaker. Existing generative models developed for AVSE do not take into account the sequential nature of speech data, which prevents them from fully incorporating the power of visual data. In this paper, we present an audiovisual deep Kalman filter (AV-DKF) generative model which assumes a first-order Markov chain model for the latent variables and effectively fuses audiovisual data. Moreover, we develop an efficient inference methodology to estimate speech signals at test time. We conduct a set of experiments to compare different variants of generative models for speech enhancement. The results demonstrate the superiority of the AV-DKF model compared with both its audio-only version and the non-sequential audio-only and audiovisual VAE-based models.