Fewer Hallucinations, More Verification: A Three-Stage LLM-Based Framework for ASR Error Correction

Re-assign community