SpecASR: Accelerating LLM-based Automatic Speech Recognition via Speculative Decoding

Re-assign community