Learning When to Speak: Latency and Quality Trade-offs for Simultaneous Speech-to-Speech Translation with Offline Models
This addresses latency-sensitive real-world applications for users needing immediate translations, though it is incremental as it builds on existing methods.
The paper tackles the problem of high latency in speech-to-speech translation by introducing a simultaneous system with tunable parameters and policies, achieving offline-level accuracy with minimal latency increases over a baseline.
Recent work in speech-to-speech translation (S2ST) has focused primarily on offline settings, where the full input utterance is available before any output is given. This, however, is not reasonable in many real-world scenarios. In latency-sensitive applications, rather than waiting for the full utterance, translations should be spoken as soon as the information in the input is present. In this work, we introduce a system for simultaneous S2ST targeting real-world use cases. Our system supports translation from 57 languages to English with tunable parameters for dynamically adjusting the latency of the output -- including four policies for determining when to speak an output sequence. We show that these policies achieve offline-level accuracy with minimal increases in latency over a Greedy (wait-$k$) baseline. We open-source our evaluation code and interactive test script to aid future SimulS2ST research and application development.