Demucs Mdx Extra

Demucs-mdx_extra provides pretrained model weights for the Demucs system, which is used for music source separation in the waveform domain. The model card indicates that this resource is a collection of four separate models, with their outputs averaged as described in the mdx_extra.yaml configuration. Each model's weights are stored in a .safetensors file, which includes metadata about the model class and initialization arguments in JSON format, and additional training metadata in an accompanying .json sidecar file. The tool is distributed under the MIT license, making it available for open source use. The evidence does not specify a particular target audience, but the context suggests it is intended for those working with music source separation, possibly including researchers and developers. Instructions for using these weights are referenced in the Demucs repository, though the evidence does not detail the usage process itself. Demucs-mdx_extra is hosted on Hugging Face, but there is no information about deployment via inference providers or tracked downloads. The evidence does not mention any pricing, so only the open source MIT license can be confirmed. No further details about integrations, supported platforms, or additional features are provided.