MuScriptor
muscriptor.kyutai.org
- Free plan
- Open source
- Windows
- macOS
- Linux
- Web
- API
- Self-hosted
What MuScriptor is
MuScriptor turns audio recordings into MIDI and written music. Its hosted service provides downloadable MIDI, sheet-music PDFs, and guitar-tab PDFs, while the web demo accepts MP3, WAV, FLAC, OGG, M4A, AIFF, and Opus files. It can transcribe voice and instruments including piano, guitar, bass, strings, brass, woodwinds, and drums; users can specify expected instruments to guide the transcription. Guitar and bass output can include tablature alongside standard notation, but chord symbols are not produced. MusicXML output can be edited in MuseScore, Sibelius, or Guitar Pro. Users can self-host the same web interface, or transcribe through a command-line interface and Python API. Local use requires Hugging Face authentication and acceptance of the CC BY-NC 4.0 model license. Hosted audio is processed on the server and not retained afterwards; local operation keeps audio on the user's machine. MuScriptor says it performs better on acoustic and guitar music than electronic music, and that results are not perfect.
Who it suits
MuScriptor suits musicians who want to turn recordings into MIDI, notation, or guitar and bass tablature. It may also suit developers or users who can meet the authentication and license requirements for local use.
Good to know, and what to watch
- Hosted service provides MIDI and PDF downloads.
- Accepts seven listed audio formats.
- Users can specify expected instruments.
- Offers a CLI and Python API.
- Local use keeps audio on the user's machine.
- Chord symbols are not produced.
- Results are less effective on electronic music.
- Local use requires authentication and license acceptance.
Plans
Every plan and priceFree
Free open-source MuScriptor
Free plan
Getting started
- Open the MuScriptor website
- Upload an MP3, WAV, FLAC, OGG, M4A, AIFF, or Opus file to the web demo
- Specify the expected instruments when appropriate
- Download the resulting MIDI, sheet-music PDF, or guitar-tab PDF
- For local use, authenticate with Hugging Face and accept the CC BY-NC 4.0 model license
- Self-host the web interface with `uvx muscriptor serve`, or use the CLI or Python API
Limits to check
The free hosted plan is listed as free. Transcription is not perfect, does not include chord symbols, and the service reports better results for acoustic and guitar music than electronic music.
The verdict
MuScriptor offers several transcription outputs and hosted or local workflows. Expect to review results, especially for electronic music, and check the model license before local use.
Questions people ask
How much does MuScriptor cost?
The Free plan is listed as free and includes the hosted demo, MIT-licensed code, and model weights under CC BY-NC 4.0 for non-commercial use.
Which audio formats can I upload?
The web demo accepts MP3, WAV, FLAC, OGG, M4A, AIFF, and Opus files.
Can it create guitar tabs?
Yes. Fretted instruments such as guitar and bass can produce tablature alongside standard notation.
Does MuScriptor generate chord symbols?
No. The transcription system does not write chord symbols.
Can I run MuScriptor locally?
Yes. The web interface can be self-hosted with `uvx muscriptor serve`, and transcription is available through a CLI and Python API. Local use requires Hugging Face authentication and acceptance of the CC BY-NC 4.0 model license.
What happens to uploaded audio?
Hosted audio is transcribed on the server and not kept afterwards. Local operation keeps audio on the user’s machine.
What the maker publishes
- Accuracy limitation
- The service says it works better on acoustic and guitar music than electronic music and is not perfect. source
- API detail
- The Python API supports instrument filtering, sampling, temperature, beam search, batching and prelude forcing. source
- Audio formats
- The web demo accepts MP3, WAV, FLAC, OGG, M4A, AIFF and Opus files. source
- Browser interface
- The hosted site runs as a JavaScript web transcriber with no installation described on its home page. source
- CLI
- The package provides a command-line transcribe command that writes MIDI, JSON, JSONL or sheet outputs. source
- Code licence
- The inference and application code is released under the MIT licence. source
- Crunchbase facts
- Crunchbase lists Kyutai as an active Paris nonprofit with 11–50 employees and founders Laurent Mazare and Neil Zeghidour. source
- Documentation depth
- Documentation includes installation, model selection, CLI examples, Python usage, server operation, output formats and development setup. source
- Full-mix transcription
- The model is designed to transcribe complete mixes containing several simultaneous instruments. source
- Hardware guidance
- Small is recommended for CPU-only machines, medium is the default trade-off, and large is intended for GPU use; Apple Silicon uses Metal automatically. source
- Headquarters
- Paris, France source
- Independent technical coverage
- A Japanese independent article reports that MuScriptor handles full mixes, explains its training approach and compares local, browser and Mirelo Studio use. source
- Independent usage report
- A Russian video review demonstrates local installation, memory use, MIDI export and opening results in MIDI editors. source
- Input formats
- The local package accepts WAV and other formats supported by libsndfile, including MP3, FLAC, OGG and M4A. source
- Instrument detection
- It detects and labels parts such as voice, drums, bass and synth lead. source
- Instruments
- It can transcribe voice, piano, guitars, bass, strings, brass, woodwinds, drums and other instruments. source
- Instrument targeting
- Users can specify expected instruments to improve transcription accuracy. source
- Integrations
- MusicXML output can be edited in MuseScore, Sibelius or Guitar Pro. source
- Kyutai company
- LinkedIn lists Kyutai as a Paris nonprofit founded in 2023 with 11–50 employees. source
- Kyutai funding
- Kyutai says it is funded by Iliad Group, CMA CGM Group and Schmidt Sciences. source
- Launch news
- Kyutai and Mirelo announced the open model and free audio-to-MIDI product on 9 July 2026. source
- Limit: local performance
- A reproducibility report found Large inference on an RTX 4090 Docker setup over six times slower than manual hosted-demo timings. source
- Limit: MusicXML
- A Windows user opened a bug stating that generated MusicXML differed substantially from the source score. source
- Limit: transcription errors
- A reported case had sustained blocks, overlapping instruments, wrong instrument labels and a slowdown toward the end of a song. source
Checked against the maker's pages on 14 Oct 2026.