Adversarial Training for Multi-Channel Sign Language Production

TL;DR

This paper introduces an adversarial multi-channel approach to Sign Language Production, utilizing a transformer-based generator and discriminator to produce realistic manual and non-manual sign features, achieving state-of-the-art results.

Contribution

It presents a novel adversarial framework that jointly models manual and non-manual sign features for more realistic sign language synthesis.

Findings

Achieved state-of-the-art back-translation performance on PHOENIX14T dataset.

Set new benchmarks for multi-channel sign language production.

Demonstrated the effectiveness of adversarial training in sign language synthesis.

Abstract

Sign Languages are rich multi-channel languages, requiring articulation of both manual (hands) and non-manual (face and body) features in a precise, intricate manner. Sign Language Production (SLP), the automatic translation from spoken to sign languages, must embody this full sign morphology to be truly understandable by the Deaf community. Previous work has mainly focused on manual feature production, with an under-articulated output caused by regression to the mean. In this paper, we propose an Adversarial Multi-Channel approach to SLP. We frame sign production as a minimax game between a transformer-based Generator and a conditional Discriminator. Our adversarial discriminator evaluates the realism of sign production conditioned on the source text, pushing the generator towards a realistic and articulate output. Additionally, we fully encapsulate sign articulators with the inclusion of non-manual features, producing facial features and mouthing patterns. We evaluate on the challenging RWTH-PHOENIX-Weather-2014T (PHOENIX14T) dataset, and report state-of-the art SLP back-translation performance for manual production. We set new benchmarks for the production of multi-channel sign to underpin future research into realistic SLP.