A multilingual speech recognizer, trained by distilling language-specialized reinforcement learning teachers into one student, outperforms each teacher on almost every tested benchmark.
T eacher 1 supervision T eacher Model m ( Encoder + adaptor ) T eacher Model m (LLM decoder) T oken 1 T oken 2 T oken n
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Language-Specialized Multi-Teacher On-Policy Distillation for Multilingual LLM-Based ASR
A multilingual speech recognizer, trained by distilling language-specialized reinforcement learning teachers into one student, outperforms each teacher on almost every tested benchmark.