FastLTS: Non-Autoregressive End-to-End Unconstrained Lip-to-Speech Synthesis

TL;DR

FastLTS is a novel non-autoregressive end-to-end lip-to-speech synthesis model that directly generates high-quality speech from videos, significantly improving speed and quality over existing methods.

Contribution

The paper introduces FastLTS, a non-autoregressive, end-to-end model with a transformer-based visual frontend for unconstrained lip-to-speech synthesis, reducing latency and memory usage.

Findings

Achieves 19.76x faster waveform generation than autoregressive models.

Produces superior speech quality in unconstrained lip-to-speech tasks.

First to use transformer-based visual frontend for this application.

Abstract

Unconstrained lip-to-speech synthesis aims to generate corresponding speeches from silent videos of talking faces with no restriction on head poses or vocabulary. Current works mainly use sequence-to-sequence models to solve this problem, either in an autoregressive architecture or a flow-based non-autoregressive architecture. However, these models suffer from several drawbacks: 1) Instead of directly generating audios, they use a two-stage pipeline that first generates mel-spectrograms and then reconstructs audios from the spectrograms. This causes cumbersome deployment and degradation of speech quality due to error propagation; 2) The audio reconstruction algorithm used by these models limits the inference speed and audio quality, while neural vocoders are not available for these models since their output spectrograms are not accurate enough; 3) The autoregressive model suffers from high inference latency, while the flow-based model has high memory occupancy: neither of them is efficient enough in both time and memory usage. To tackle these problems, we propose FastLTS, a non-autoregressive end-to-end model which can directly synthesize high-quality speech audios from unconstrained talking videos with low latency, and has a relatively small model size. Besides, different from the widely used 3D-CNN visual frontend for lip movement encoding, we for the first time propose a transformer-based visual frontend for this task. Experiments show that our model achieves $19.76\times$ speedup for audio waveform generation compared with the current autoregressive model on input sequences of 3 seconds, and obtains superior audio quality.