SLAM-LLM

Kai Yu's X-LANCE modular training framework for speech, audio, and music LLMs (1K★). Home of SLAM-ASR (arXiv:2402.08846) — the "embarrassingly simple" linear-projector speech-recognition LLM recipe — and SLAM-Omni (Dec 2024), a single-stage-trained timbre-controllable voice-interaction model.