StreamSpeech: Simultaneous Speech-to-Speech Translation with Multi-task Learning

TL;DR

StreamSpeech is a unified multi-task learning model that performs simultaneous speech translation, recognition, and synthesis, achieving state-of-the-art results and providing high-quality intermediate outputs for real-time communication.

Contribution

It introduces a novel unified model that jointly learns translation and policy for simultaneous speech translation within a multi-task framework.

Findings

Achieves state-of-the-art performance on CVSS benchmark

Provides high-quality intermediate recognition and translation results

Supports offline and real-time speech translation, recognition, and synthesis

Abstract

Simultaneous speech-to-speech translation (Simul-S2ST, a.k.a streaming speech translation) outputs target speech while receiving streaming speech inputs, which is critical for real-time communication. Beyond accomplishing translation between speech, Simul-S2ST requires a policy to control the model to generate corresponding target speech at the opportune moment within speech inputs, thereby posing a double challenge of translation and policy. In this paper, we propose StreamSpeech, a direct Simul-S2ST model that jointly learns translation and simultaneous policy in a unified framework of multi-task learning. Adhering to a multi-task learning approach, StreamSpeech can perform offline and simultaneous speech recognition, speech translation and speech synthesis via an "All-in-One" seamless model. Experiments on CVSS benchmark demonstrate that StreamSpeech achieves state-of-the-art performance in both offline S2ST and Simul-S2ST tasks. Besides, StreamSpeech is able to present high-quality intermediate results (i.e., ASR or translation results) during simultaneous translation process, offering a more comprehensive real-time communication experience.