SimulU: Training-free Policy for Long-form Simultaneous Speech-to-Speech Translation

TL;DR

SimulU introduces a training-free policy for long-form simultaneous speech-to-speech translation, leveraging pre-trained models to improve real-time multilingual communication without extensive training.

Contribution

It is the first approach to enable training-free, long-form SimulS2S using history management and cross-attention strategies in pre-trained models.

Findings

Achieves comparable quality-latency trade-offs to trained models.

Operates effectively on continuous, long-form speech.

Eliminates need for resource-intensive training procedures.

Abstract

Simultaneous speech-to-speech translation (SimulS2S) is essential for real-time multilingual communication, with increasing integration into meeting and streaming platforms. Despite this, SimulS2S remains underexplored in research, where current solutions often rely on resource-intensive training procedures and operate on short-form, pre-segmented utterances, failing to generalize to continuous speech. To bridge this gap, we propose SimulU, the first training-free policy for long-form SimulS2S. SimulU adopts history management and speech output selection strategies that exploit cross-attention in pre-trained end-to-end models to regulate both input history and output generation. Evaluations on MuST-C across 8 languages show that SimulU achieves a better or comparable quality-latency trade-off against strong cascaded models. By eliminating the need for ad-hoc training, SimulU offers a promising path to end-to-end SimulS2S in realistic, long-form scenarios.