Align, Write, Re-order: Explainable End-to-End Speech Translation via Operation Sequence Generation

TL;DR

This paper introduces an explainable end-to-end speech translation method that generates aligned, re-orderable translations by predicting operation sequences, improving interpretability and performance in streaming and offline settings.

Contribution

It proposes a novel operation sequence generation approach that enables explicit alignment and reordering in end-to-end speech translation systems, enhancing explainability and efficiency.

Findings

Provides explainable translation with explicit source-target word alignment.

Improves streaming translation performance through delayed re-ordering.

Achieves faster combined ASR and ST by joint prediction.

Abstract

The black-box nature of end-to-end speech translation (E2E ST) systems makes it difficult to understand how source language inputs are being mapped to the target language. To solve this problem, we would like to simultaneously generate automatic speech recognition (ASR) and ST predictions such that each source language word is explicitly mapped to a target language word. A major challenge arises from the fact that translation is a non-monotonic sequence transduction task due to word ordering differences between languages -- this clashes with the monotonic nature of ASR. Therefore, we propose to generate ST tokens out-of-order while remembering how to re-order them later. We achieve this by predicting a sequence of tuples consisting of a source word, the corresponding target words, and post-editing operations dictating the correct insertion points for the target word. We examine two variants of such operation sequences which enable generation of monotonic transcriptions and non-monotonic translations from the same speech input simultaneously. We apply our approach to offline and real-time streaming models, demonstrating that we can provide explainable translations without sacrificing quality or latency. In fact, the delayed re-ordering ability of our approach improves performance during streaming. As an added benefit, our method performs ASR and ST simultaneously, making it faster than using two separate systems to perform these tasks.