Skip to content
/ jams Public

A JSON Annotated Music Specification for Reproducible MIR Research

License

Notifications You must be signed in to change notification settings

marl/jams

Folders and files

NameName
Last commit message
Last commit date

Latest commit

08fb73f · Feb 11, 2017
Feb 11, 2017
Feb 11, 2017
Jan 31, 2016
Feb 11, 2017
Feb 4, 2016
Feb 29, 2016
Feb 11, 2017
Sep 2, 2016
Aug 19, 2015
Jun 30, 2016
Jul 12, 2016
Aug 29, 2016

Repository files navigation

jams

PyPI License Build Status Coverage Status Dependency Status

A JSON Annotated Music Specification for Reproducible MIR Research.

Please, refer to documentation for a comprehensive description of JAMS.

What

JAMS is a JSON-based music annotation format.

We provide:

  • A formal JSON schema for generic annotations
  • The ability to store multiple annotations per file
  • Schema definitions for a wide range of annotation types (beats, chords, segments, tags, etc.)
  • Error detection and validation for annotations
  • A translation layer to interface with mir eval for evaluating annotations

Why

Music annotations are traditionally provided as plain-text files employing simple formatting schema (comma or tab separated) when possible. However, as the field of MIR has continued to evolve, such annotations have become increasingly complex, and more often custom conventions are employed to represent this information. And, as a result, these custom conventions can be unwieldy and non-trivial to parse and use.

Therefore, JAMS provides a simple, structured, and sustainable approach to representing rich information in a human-readable, language agnostic format. Importantly, JAMS supports the following use-cases:

  • multiple types annotations
  • multiple annotations for a given task
  • rich file level and annotation level metadata

How

This library is offered as a proof-of-concept, demonstrating the promise of a JSON-based schema to meet the needs of the MIR community. To install, clone the repository into a working directory and proceed thusly.

The full documentation can be found here.

Who

To date, the initial JAMS effort has evolved out of internal needs at MARL@NYU, with some great feedback from our friends at LabROSA.

If you want to get involved, do let us know!

Details

JAMS is proposed in the following publication:

Eric J. Humphrey, Justin Salamon, Oriol Nieto, Jon Forsyth, Rachel M. Bittner, and Juan P. Bello, "JAMS: A JSON Annotated Music Specification for Reproducible MIR Research", Proceedings of the 15th International Conference on Music Information Retrieval, 2014.