Code-Switched Language Models Using Neural Based Synthetic Data from Parallel Sentences

TL;DR

This paper introduces a neural sequence-to-sequence model with a copy mechanism that generates synthetic code-switching data from parallel monolingual sentences, improving language models and speech recognition without external linguistic tools.

Contribution

It presents a novel neural model that learns to generate realistic code-switching sentences using only parallel data, avoiding external alignment tools and capturing linguistic constraints.

Findings

Achieves state-of-the-art language model performance.

Enhances end-to-end speech recognition accuracy.

Generates high-quality synthetic code-switching data.

Abstract

Training code-switched language models is difficult due to lack of data and complexity in the grammatical structure. Linguistic constraint theories have been used for decades to generate artificial code-switching sentences to cope with this issue. However, this require external word alignments or constituency parsers that create erroneous results on distant languages. We propose a sequence-to-sequence model using a copy mechanism to generate code-switching data by leveraging parallel monolingual translations from a limited source of code-switching data. The model learns how to combine words from parallel sentences and identifies when to switch one language to the other. Moreover, it captures code-switching constraints by attending and aligning the words in inputs, without requiring any external knowledge. Based on experimental results, the language model trained with the generated sentences achieves state-of-the-art performance and improves end-to-end automatic speech recognition.