Segmentation-Free Streaming Machine Translation

TL;DR

This paper introduces a segmentation-free streaming machine translation framework that translates unsegmented input streams in real-time, improving quality and latency over traditional segmentation-dependent methods.

Contribution

It proposes a novel segmentation-free approach that delays segmentation until after translation, enhancing real-time translation quality and latency trade-offs.

Findings

Outperforms segmentation-based methods in quality-latency trade-off

Enables translation of unsegmented streams without intermediate segmentation

Shows significant improvements in real-time translation performance

Abstract

Streaming Machine Translation (MT) is the task of translating an unbounded input text stream in real-time. The traditional cascade approach, which combines an Automatic Speech Recognition (ASR) and an MT system, relies on an intermediate segmentation step which splits the transcription stream into sentence-like units. However, the incorporation of a hard segmentation constrains the MT system and is a source of errors. This paper proposes a Segmentation-Free framework that enables the model to translate an unsegmented source stream by delaying the segmentation decision until the translation has been generated. Extensive experiments show how the proposed Segmentation-Free framework has better quality-latency trade-off than competing approaches that use an independent segmentation model. Software, data and models will be released upon paper acceptance.