Skip to main content

Manipulate your MIDI file in bar level, and converting between MIDI and REMI-z format.

Project description

REMI-z tokenizer and MultiTrack music data structure

This is the official implementation of the REMI-z tokenizer in the paper Unlocking Potential in Pre-Trained Music Language Models for Versatile Multi-Track Music Arrangement.

[github | PyPI | author]

This tool helps to convert your music between MIDI and REMI-z representation, which is an efficient sequence representation of multitrack music, meanwhile facilitate manipulate the music at bar level.

The core of this tokenizer is the MultiTrack class as the data structure for multitrack music, which is a hierachical format. Here are the structural details:

  • The music is represented by an MultiTrack object, which is list of bars.
    • Each Bar object represents all notes being played within one bar, grouped by Track object, together with time signature and tempo info of this bar.
      • Each Track object represents one instrument, contatining notes of that instrument in this bar.
        • Each Note object represent one note, including onset, offset, pitch, velocity information.

This Multitrack object can be create from various formats (e.g., MIDI or REMI-z), and convert into various formats (e.g., MIDI, and REMI-z representation).

Install

Install from pip

pip install REMI-z

Install from source

git clone https://github.com/Sonata165/REMI-z.git
cd REMI-z
pip install -r requirements.txt
pip install .

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

remi_z-0.5.7.tar.gz (4.1 kB view details)

Uploaded Source

File details

Details for the file remi_z-0.5.7.tar.gz.

File metadata

  • Download URL: remi_z-0.5.7.tar.gz
  • Upload date:
  • Size: 4.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.0.1 CPython/3.12.7

File hashes

Hashes for remi_z-0.5.7.tar.gz
Algorithm Hash digest
SHA256 ed10efb6de1de4f0bdc3eaff269b3750dc5e28414578e25bcd0562a1a3ebfac5
MD5 bea252a2372e6772579d85c1dee65bfb
BLAKE2b-256 49a81ac0923bd9c8603e57718591211d9415001510c08169b5a486ff6de1f6f3

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page