SimulSLT: End-to-End Simultaneous Sign Language Translation

TL;DR

SimulSLT is an innovative end-to-end model for simultaneous sign language translation that reduces latency and improves translation quality by using a novel boundary predictor and re-encode method, enabling real-time applications.

Contribution

The paper introduces SimulSLT, the first end-to-end simultaneous sign language translation model with a boundary predictor and re-encode technique for improved real-time translation.

Findings

Achieves higher BLEU scores than non-simultaneous models.

Maintains low latency suitable for real-time use.

Effective boundary prediction enhances translation accuracy.

Abstract

Sign language translation as a kind of technology with profound social significance has attracted growing researchers' interest in recent years. However, the existing sign language translation methods need to read all the videos before starting the translation, which leads to a high inference latency and also limits their application in real-life scenarios. To solve this problem, we propose SimulSLT, the first end-to-end simultaneous sign language translation model, which can translate sign language videos into target text concurrently. SimulSLT is composed of a text decoder, a boundary predictor, and a masked encoder. We 1) use the wait-k strategy for simultaneous translation. 2) design a novel boundary predictor based on the integrate-and-fire module to output the gloss boundary, which is used to model the correspondence between the sign language video and the gloss. 3) propose an innovative re-encode method to help the model obtain more abundant contextual information, which allows the existing video features to interact fully. The experimental results conducted on the RWTH-PHOENIX-Weather 2014T dataset show that SimulSLT achieves BLEU scores that exceed the latest end-to-end non-simultaneous sign language translation model while maintaining low latency, which proves the effectiveness of our method.