SpeechAlign: a Framework for Speech Translation Alignment Evaluation

TL;DR

SpeechAlign introduces a new framework with datasets and metrics for evaluating source-target alignment in speech translation models, facilitating progress in speech-to-speech and speech-to-text translation research.

Contribution

It provides the first dedicated evaluation dataset and novel metrics for alignment quality in speech translation models, enabling standardized benchmarking.

Findings

Introduced Speech Gold Alignment dataset for English-German translation

Proposed two new metrics: SAER and TW-SAER for alignment evaluation

Benchmarking of open-source speech translation models using SpeechAlign

Abstract

Speech-to-Speech and Speech-to-Text translation are currently dynamic areas of research. In our commitment to advance these fields, we present SpeechAlign, a framework designed to evaluate the underexplored field of source-target alignment in speech models. The SpeechAlign framework has two core components. First, to tackle the absence of suitable evaluation datasets, we introduce the Speech Gold Alignment dataset, built upon a English-German text translation gold alignment dataset. Secondly, we introduce two novel metrics, Speech Alignment Error Rate (SAER) and Time-weighted Speech Alignment Error Rate (TW-SAER), which enable the evaluation of alignment quality within speech models. While the former gives equal importance to each word, the latter assigns weights based on the length of the words in the speech signal. By publishing SpeechAlign we provide an accessible evaluation framework for model assessment, and we employ it to benchmark open-source Speech Translation models. In doing so, we contribute to the ongoing research progress within the fields of Speech-to-Speech and Speech-to-Text translation.