Neural Language Codes for Multilingual Acoustic Models

TL;DR

This paper introduces a novel multilingual acoustic model that uses language codes generated by an auxiliary network to adapt quickly to different languages and accents without retraining, outperforming monolingual models.

Contribution

The paper proposes a Meta-Pi network architecture that modulates acoustic models with language codes, enabling rapid adaptation to multiple languages without additional training or data.

Findings

Meta-Pi networks adapt quickly to new languages during recognition

Multilingual models outperform monolingual models in recognition accuracy

Language codes effectively encode language-specific variations

Abstract

Multilingual Speech Recognition is one of the most costly AI problems, because each language (7,000+) and even different accents require their own acoustic models to obtain best recognition performance. Even though they all use the same phoneme symbols, each language and accent imposes its own coloring or "twang". Many adaptive approaches have been proposed, but they require further training, additional data and generally are inferior to monolingually trained models. In this paper, we propose a different approach that uses a large multilingual model that is \emph{modulated} by the codes generated by an ancillary network that learns to code useful differences between the "twangs" or human language. We use Meta-Pi networks to have one network (the language code net) gate the activity of neurons in another (the acoustic model nets). Our results show that during recognition multilingual Meta-Pi networks quickly adapt to the proper language coloring without retraining or new data, and perform better than monolingually trained networks. The model was evaluated by training acoustic modeling nets and modulating language code nets jointly and optimize them for best recognition performance.