None defined yet.
Batched Speech Decisions Without Decoding: Single-Token Supervision Lets a Frozen LLM Hear Beyond the Transcript
Ten decisions from raw speech in one forward pass
