Engine
Do you plan to develop a minimax engine like the one you did on AceStep which has been great so far- my no. 1 for us with older/ smaller VRAM GPUs
Yes it's done : https://github.com/ServeurpersoCom/minimaxmusic.cpp
The model has far fewer features (cover, repaint, etc.) than the one from acestep.cpp, but it offers much greater diversity outside of electronic music. It also consumes more resources and slightly more memory. You can test the application and the quantizations; by default, the strategy prioritizes memory efficiency (RAII).
I’ve incorporated the author’s example prompts; I highly recommend using a large external LLM to generate them, there’s even a skill available in the upstream project, though simply copying and pasting into llama.cpp Qwen3.8 or the free version of ChatGPT works just fine!
Testing it tonite this looks great so far. Would recommend it for any one with smaller GPUs. Great job Thanks man