Vision as an Interlingua: Learning Multilingual Semantic Embeddings of Untranscribed Speech

TL;DR

This paper introduces a neural network approach to learn multilingual semantic embeddings directly from speech waveforms in English and Hindi, enabling cross-lingual retrieval without transcriptions.

Contribution

It is the first to extend multilingual speech embedding learning to languages beyond English, demonstrating improved performance and cross-lingual capabilities.

Findings

Multilingual models outperform monolingual models.

Models successfully perform cross-lingual speech-to-speech retrieval.

Embeddings are learned directly from waveforms without transcriptions.

Abstract

In this paper, we explore the learning of neural network embeddings for natural images and speech waveforms describing the content of those images. These embeddings are learned directly from the waveforms without the use of linguistic transcriptions or conventional speech recognition technology. While prior work has investigated this setting in the monolingual case using English speech data, this work represents the first effort to apply these techniques to languages beyond English. Using spoken captions collected in English and Hindi, we show that the same model architecture can be successfully applied to both languages. Further, we demonstrate that training a multilingual model simultaneously on both languages offers improved performance over the monolingual models. Finally, we show that these models are capable of performing semantic cross-lingual speech-to-speech retrieval.