Research code for continual multilingual ASR and code-switching speech recognition with Whisper, Qwen2-Audio, LoRA, Bayesian low-rank factorization, and weight centralization.
pytorch speech-recognition automatic-speech-recognition speech-processing asr low-rank-factorization bayesian-learning continual-learning code-switching multilingual-speech-recognition weight-factorization language-balance model-adaptation code-switching-speech robust-adaptation
-
Updated
Jun 21, 2026 - Python