Tackling data scarcity in speech translation using zero-shot multilingual machine translation techniques

TL;DR

This paper explores zero-shot multilingual techniques to improve speech translation performance under data scarcity by leveraging speech transcription and text translation data, with promising results in low-resource scenarios.

Contribution

It adapts zero-shot translation ideas from text to speech translation, demonstrating effective data augmentation and auxiliary loss functions for low-resource speech translation.

Findings

Achieved up to +12.9 BLEU points in low-resource settings.

Improved over direct end-to-end speech translation models.

Enhanced performance with auxiliary loss and data augmentation techniques.

Abstract

Recently, end-to-end speech translation (ST) has gained significant attention as it avoids error propagation. However, the approach suffers from data scarcity. It heavily depends on direct ST data and is less efficient in making use of speech transcription and text translation data, which is often more easily available. In the related field of multilingual text translation, several techniques have been proposed for zero-shot translation. A main idea is to increase the similarity of semantically similar sentences in different languages. We investigate whether these ideas can be applied to speech translation, by building ST models trained on speech transcription and text translation data. We investigate the effects of data augmentation and auxiliary loss function. The techniques were successfully applied to few-shot ST using limited ST data, with improvements of up to +12.9 BLEU points compared to direct end-to-end ST and +3.1 BLEU points compared to ST models fine-tuned from ASR model.