https://github.com/magenta/mt3

MT3: Multi-Task Multitrack Music Transcription
https://github.com/magenta/mt3

Last synced: 1 day ago
JSON representation

MT3: Multi-Task Multitrack Music Transcription

• Host: GitHub
• URL: https://github.com/magenta/mt3
• Owner: magenta
• License: apache-2.0
• Created: 2021-11-03T22:15:44.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2024-12-11T03:42:32.000Z (about 1 month ago)
• Last Synced: 2025-01-02T20:04:25.976Z (9 days ago)
• Language: Python
• Size: 123 KB
• Stars: 1,458
• Watchers: 26
• Forks: 195
• Open Issues: 46
• Metadata Files:
  • Readme: README.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE

Awesome Lists containing this project

• my-awesome - magenta/mt3 - 12 star:1.5k fork:0.2k MT3: Multi-Task Multitrack Music Transcription (Python)

README

        
# MT3: Multi-Task Multitrack Music Transcription

MT3 is a multi-instrument automatic music transcription model that uses the [T5X framework](https://github.com/google-research/t5x).

This is not an officially supported Google product.

## Transcribe your own audio

Use our [colab notebook](https://colab.research.google.com/github/magenta/mt3/blob/main/mt3/colab/music_transcription_with_transformers.ipynb) to

transcribe audio files of your choosing.  You can use a pretrained checkpoint from

either a) the piano transcription model described in [our ISMIR 2021 paper](https://archives.ismir.net/ismir2021/paper/000030.pdf)

or b) the multi-instrument transcription model described in

[our ICLR 2022 paper](https://openreview.net/pdf?id=iMSjopcOn0p).

## Train a model

For now, we do not (easily) support training.  If you like, you can try to

follow the [T5X training instructions](https://github.com/google-research/t5x#training)

and use one of the tasks defined in [tasks.py](mt3/tasks.py).