AlloST: Low-resource Speech Translation without Source Transcription

TL;DR

This paper introduces a low-resource speech translation framework that leverages a universal phone recognizer and phonetic embeddings, significantly improving translation quality without relying on source transcriptions.

Contribution

It proposes a novel attention-based sequence-to-sequence model utilizing phonetic embeddings and BPE segmentation, advancing low-resource speech translation without source transcription.

Findings

Outperforms conformer-based baseline models

Achieves performance close to methods using source transcription

Effective on Spanish-English and Mandarin dialect corpora

Abstract

The end-to-end architecture has made promising progress in speech translation (ST). However, the ST task is still challenging under low-resource conditions. Most ST models have shown unsatisfactory results, especially in the absence of word information from the source speech utterance. In this study, we survey methods to improve ST performance without using source transcription, and propose a learning framework that utilizes a language-independent universal phone recognizer. The framework is based on an attention-based sequence-to-sequence model, where the encoder generates the phonetic embeddings and phone-aware acoustic representations, and the decoder controls the fusion of the two embedding streams to produce the target token sequence. In addition to investigating different fusion strategies, we explore the specific usage of byte pair encoding (BPE), which compresses a phone sequence into a syllable-like segmented sequence. Due to the conversion of symbols, a segmented sequence represents not only pronunciation but also language-dependent information lacking in phones. Experiments conducted on the Fisher Spanish-English and Taigi-Mandarin drama corpora show that our method outperforms the conformer-based baseline, and the performance is close to that of the existing best method using source transcription.