Unsupervised Cross-Modal Alignment of Speech and Text Embedding Spaces

TL;DR

This paper introduces an unsupervised method to align speech and text embedding spaces, enabling cross-modal tasks like translation and classification, especially beneficial for low-resource languages with limited parallel data.

Contribution

The paper proposes a novel adversarial training framework for unsupervised alignment of speech and text embeddings, extending cross-lingual embedding techniques to the speech-text domain.

Findings

Achieves comparable performance to supervised methods in spoken word classification and translation.

Demonstrates effectiveness for low-resource languages with minimal parallel data.

Provides a foundation for developing ASR and speech translation systems in underrepresented languages.

Abstract

Recent research has shown that word embedding spaces learned from text corpora of different languages can be aligned without any parallel data supervision. Inspired by the success in unsupervised cross-lingual word embeddings, in this paper we target learning a cross-modal alignment between the embedding spaces of speech and text learned from corpora of their respective modalities in an unsupervised fashion. The proposed framework learns the individual speech and text embedding spaces, and attempts to align the two spaces via adversarial training, followed by a refinement procedure. We show how our framework could be used to perform spoken word classification and translation, and the results on these two tasks demonstrate that the performance of our unsupervised alignment approach is comparable to its supervised counterpart. Our framework is especially useful for developing automatic speech recognition (ASR) and speech-to-text translation systems for low- or zero-resource languages, which have little parallel audio-text data for training modern supervised ASR and speech-to-text translation models, but account for the majority of the languages spoken across the world.