MuScriptor small - ONNX export for MidiEditor AI

An ONNX conversion of MuScriptor/muscriptor-small, the multi-instrument music transcription model by Kyutai and Mirelo. It lets the audio-to-MIDI transcription of MidiEditor AI run inside the editor with ONNX Runtime, without Python.

This repository is not affiliated with Kyutai or Mirelo. All credit for the model belongs to its authors. For the original weights, the paper, the code and the hosted demo see:

What was changed

The weights are the original's (revision 8c127f603b807520fa465c838e9bfee8a91ada4e); the model was not retrained or fine-tuned. Changes made for this export:

  • The network was rebuilt as two ONNX graphs from model.safetensors: frontend.onnx (one 5 s chunk of 16 kHz audio to the conditioning prefix: log-mel frames through the model's projection, then the class rows) and decoder.onnx with decoder.onnx.data (the transformer decoder).
  • The decoder's weights and its key/value cache are stored as float16 (the original is float32); the frontend stays float32.
  • Attention uses ONNX Runtime's GroupQueryAttention operator (domain com.microsoft) with one key/value buffer per layer, so the files need ONNX Runtime and do not load in other ONNX runtimes.
  • muscriptor.json describes the export (sizes, token ids, the conditioning rows).

Files

File Size What
frontend.onnx 6.5 MB audio chunk to conditioning prefix
decoder.onnx 35 KB the decoder's graph
decoder.onnx.data 205 MB the decoder's weights (float16)
muscriptor.json 1 KB description of the export

Use

In MidiEditor AI: MidiCreator > Transcribe > "Models...". The editor downloads the four files into one folder. To use them elsewhere, keep all four in the same folder.

License and conditions of use

The weights are released by their authors under CC BY-NC 4.0 - non-commercial use only - supplemented by the specific conditions of use below. This export is an adapted version of those weights and is shared under the same license and the same conditions. By downloading or using these files you accept them.

The authors' statement, quoted unchanged from the original model's page:

MuScriptor is the result of a research collaboration between Mirelo and Kyutai whose purpose is to transcribe audio to MIDI/music sheet. It is provided primarily for research purposes under the CC BY-NC 4.0 licence supplemented by the below specific conditions of use.

Specific conditions of use: MuScriptor and any generated content by MuScriptor are provided as is without any warranty of any kind, including but not limited to any warranty of non-infringement. Use of MuScriptor and its output must comply with all applicable laws and must not result in, involve, or facilitate any illegal or unauthorized activity. Prohibited uses include, without limitation, inputting music files and transcribing them to MIDI/music sheet without having all the necessary rights, including intellectual property rights, under applicable laws. Accordingly, users of MuScriptor undertake and warrant to have all the necessary rights, including intellectual property rights, in connection with their use of MuScriptor and its output. We disclaim all liability for any non-compliant use and users of MuScriptor shall indemnify, defend, and hold harmless Mirelo and Kyutai from and against any and all claims, damages, losses, liabilities, and expenses (including reasonable attorneys' fees) incurred by Mirelo and/or Kyutai arising out of or resulting from their failure to comply with the terms of the CC BY-NC 4.0 licence and/or these specific conditions of use.

The conversion itself is provided as is, without any warranty.

Citation

@misc{rouard2026muscriptoropenmodelmultiinstrument,
      title={MuScriptor: An Open Model for Multi-Instrument Music Transcription},
      author={Simon Rouard and Michael Krause and Axel Roebel and Carl-Johann Simon-Gabriel and Alexandre Défossez},
      year={2026},
      eprint={2607.08168},
      archivePrefix={arXiv},
      primaryClass={cs.SD},
      url={https://arxiv.org/abs/2607.08168},
}
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for happytunesai/muscriptor-small-onnx

Quantized
(2)
this model

Paper for happytunesai/muscriptor-small-onnx