Towards Truly Multilingual ASR: Generalizing Code-Switching ASR to Unseen Language Pairs
It addresses the scalability bottleneck of code-switching ASR for multilingual systems, but results are incremental.
The paper investigates whether code-switching ASR capabilities learned from a limited set of language pairs can generalize to unseen pairs via model merging and domain generalization, finding only modest generalization.
Automatic Speech Recognition (ASR) has become a key technology for human--AI interaction. However, code-switching ASR (CS-ASR) remains particularly challenging due to the severe scarcity of multilingual CS speech resources across diverse language pairs. Existing approaches primarily improve CS-ASR performance through synthetic CS speech generation or pair-specific fine-tuning on limited bilingual datasets. Nevertheless, these approaches face an inherent scalability limitation, as support for CS must be developed separately for language pairs whose number grows combinatorially with the number of supported languages. In this work, we investigate whether CS capabilities learned from a limited set of seen language pairs can generalize to unseen language pairs through model merging and domain generalization methods. Our experiments show that merged bilingual CS-ASR models modestly generalize to unseen language pairs, suggesting limited transfer of bilingual CS capabilities across language pairs.