Annotations and Python tools for MedleyDB. Read the Docs here.
Maintained by Rachel Bittner rachel (dot) bittner (at) nyu (dot) edu.
This code is released along with MedleyDB and is a component of the work presented in the following publications:
R. Bittner, J. Salamon, M. Tierney, M. Mauch, C. Cannam and J. P. Bello, "MedleyDB: A Multitrack Dataset for Annotation-Intensive MIR Research", in 15th International Society for Music Information Retrieval Conference, Taipei, Taiwan, Oct. 2014.
R. Bittner, J. Wilkins, H. Yip and J. P. Bello, "MedleyDB 2.0: New Data and a System for Sustainable Data Collection", in Proceedings of the 17th International Society for Music Information Retrieval Conference Late Breaking and Demo Papers, New York City, USA, Aug. 2016.
As of v1.2, this repository contains the most up to date version of the medleydb annotations. If you find any problems with an annotation, or would like to contribute annotations, please report an issue submit a pull request. :)
First clone this repository:
bash
git clone https://github.com/marl/medleydb.git
Install the package
bash
cd medleydb
pip install .
Next, set the environment variable MEDLEYDB_PATH to the local path where the MedleyDB directory (or MedleyDB_sample) lives:
bash
export MEDLEYDB_PATH="path/to/your/copy/of/MedleyDB"
To avoid doing this step every time, copy the line above to ~/.bash_profile
or ~/.bashrc
.
If you use homebrew, you can install sox by doing:
bash
brew install sox # install the binary
pip install sox # install the python package
Known issues with the Audio and Metadata can be found here.
How to get the annotations of medleydb2.0 dataset?
Tabla Breakbeat Science - WhoIsIt in branch medleydb_v2, Phoenix_ScotchMorris_METADATA in branch medleydb_v1.2, medleydb_v2 and master
I want to get the raw audio data in wav format, but I dont know how to do that.
I noticed that there is plenty more songs in the EXTRA dataset. I also noticed that there is a download script pointing to a private Google Drive. Is it possible to gain access to this extra data?
All annotations and metadata from the initial release are now version controlled. Several annotation file name errors have been corrected.
The code has been restructured to rely primarily on the version controlled metadata, rather than the audio from the full dataset.
This works on MedleyDB's original file structure.