Training a code-switching language model with monolingual data
arXiv:1911.06003
Abstract
A lack of code-switching data complicates the training of code-switching (CS) language models. We propose an approach to train such CS language models on monolingual data only. By constraining and normalizing the output projection matrix in RNN-based language models, we bring embeddings of different languages closer to each other. Numerical and visualization results show that the proposed approaches remarkably improve the performance of CS language models trained on monolingual data. The proposed approaches are comparable or even better than training CS language models with artificially generated CS data. We additionally use unsupervised bilingual word translation to analyze whether semantically equivalent words in different languages are mapped together.
arXiv commentAccepted as an oral presentation in ICASSP 2020
References in corpus (4)
- Direct Acoustics-to-Word Models for English Conversational Speech Recognition
- Learn to Code-Switch: Data Augmentation using Copy Mechanism on Language Modeling
- Constrained Output Embeddings for End-to-End Code-Switching Speech Recognition with Only Monolingual Data
- Code-switching Sentence Generation by Generative Adversarial Networks and its Application to Data Augmentation