Voxtral ASR Streaming

[patrickvonplaten]

Nice effort on trying to focus purely on audio!

Any chance Voxtral ASR streaming could be added?

Didn't yet make it into llama.cpp: ggml-org/llama.cpp#20914

Reference: https://huggingface.co/mistralai/Voxtral-Mini-4B-Realtime-2602

[0xShug0]

@patrickvonplaten Thanks! I will add it to the todo.

[0xShug0]

@patrickvonplaten Would you like to share your use case of Voxtral ASR? The model is pretty large, so I’m not sure what it brings over smaller, lighter models.