Audio Splitter models
Hybrid Transformer Demucs (htdemucs, htdemucs_6s) by Meta AI, exported to ONNX for the in-browser Audio Splitter. The STFT and iSTFT are done outside the graph. Weights are stored as float16 and cast to float32 at load time. Original models: https://github.com/adefossez/demucs (MIT license).
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support