Improving low-resource ASR using bilingual fine-tuning with language identification: a cross-linguistic evaluation

In three linesStudy on bilingual fine-tuning for low-resource ASR across 9 language pairs. Uses language identification tokens prepended to input text. Results: bilingual fine-tuning improves performance when language ID accuracy is high; providing the token at inference mitigates low language ID performance.

Summary generated by Claude — human-verified