Switching Variational Auto-Encoders for Noise-Agnostic Audio-visual Speech Enhancement
This work provides an incremental improvement for speech enhancement systems that rely on audio-visual data, especially for scenarios where visual input quality is inconsistent.
The paper addresses the challenge of audio-visual speech enhancement in unsupervised settings, particularly when visual data quality varies. It introduces a switching variational auto-encoder (SwVAE) that dynamically combines audio-only and audio-visual VAE architectures based on visual data quality, leading to promising performance.
Recently, audio-visual speech enhancement has been tackled in the unsupervised settings based on variational auto-encoders (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, audio-visual 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.