End-to-end audio-native SLM fine-tuning & model surgery framework (Qwen3-Omni backbone) for Indic telephony speech-to-speech.
pytorch telephony speech-to-speech indic-nlp llm-training qwen3-omni snac-codec model-surgery lora-finetuning audio-slm
-
Updated
Sep 21, 2026