Description

Open Broadcast Media Audio from TV (OpenBMAT) is an open, annotated dataset for the task of music detection that contains over 27 hours of TV broadcast audio from 4 countries distributed over 1647 one-minute long excerpts. It is designed to encompass several essential features for any music detection dataset and is the first one to include annotations about the loudness of music in relation to other simultaneous non-music sounds. OpenBMAT has been cross-annotated by 3 annotators obtaining high inter-annotator agreement percentages, which validates the annotation methodology and ensures the annotations reliability.

Data Source

We have sampled the audio files from from BMAT’s private database, which temporarily stores recordings from over 2000 TV channels that this company monitors as part of its business.

./audio:

- Contains all audio files (22050 Hz, 16-bit, monophonic, WAV).

./splits:

- splits.csv: a csv file assigning each audio file to a split.
- split_balance.txt: proportion of each of the MRLE_mapping classes for each split and annotator.

./annotations/bat:

- Contains the annotations as output by the annotation tool.
- One file per annotator.
- Complete taxonomy (no mapping).

./annotations/json:

- Contains the annotations in JSON format.
  - original.json: annotations using complete taxonomy.
  - MD_mapping.json: annotations using MD mapping.
  - RMLE_mapping.json: annotations using MRLE mapping.

./annotations/tsv/original:

- Contains a separate TSV file for each audio file and annotator with the corresponding annotations using the complete taxonomy.

./annotations/tsv/MD_mapping:

- Contains a separate TSV file for each audio file and annotator with the corresponding annotations using the MD mapping.

./annotations/tsv/RMLE_mapping:

- Contains a separate TSV file for each audio file and annotator with the corresponding annotations using the MRLE mapping.

./utils.py:

- Generates tsv and json annotations from bat annotations
- Generates audio excerpts with full/partial agreement
- Loads annotations as dcase_util MetaDataContainer objects

We would appreciate scientific publications of works using OpenBMAT to refer to:

Meléndez-Catalán, Blai, Molina, Emilio, & Gómez, Emilia. (2019). Open Broadcast Media Audio from TV (OpenBMAT) (Version 1.0.0) [Data set]. Transactions of the International Society for Music Information Retrieval (TISMIR). Zenodo. http://doi.org/10.5281/zenodo.3381249

The paper describing OpenBMAT should be cited as:

Meléndez-Catalán, B., Molina, E. and Gómez, E., 2019. Open Broadcast Media Audio from TV: A Dataset of TV Broadcast Audio with Relative Music Loudness Annotations. Transactions of the International Society for Music Information Retrieval, 2(1), pp.43–51. DOI: http://doi.org/10.5334/tismir.29

 

We thank the annotators for the effort they invested in the annotation of OpenBMAT. We also thank Alex Ciurana from BMAT with whom we have discussed issues essential to the construction of this dataset. We thank Brisa Burriel for her advice on the best way to make OpenBMAT available to the research community and Sonia Espi for providing the means to do so. Finally, we thank the Catalan Industrial Doctorates Plan for funding this research.

Download OpenBMAT dataset here:

OpenBMAT

 

Conditions of use

Disclaimer

Universitat Pompeu Fabra bears no responsibility for the uses of the OpenBMAT Dataset, or for
interpretations or inferences based on these uses. Universitat Pompeu Fabra accepts no liability for
indirect, consequential or incidental damages or losses arising from use of the OpenBMAT Dataset,
or from the unavailability of, or break in access to the Dataset for whatever reason.

Universitat Pompeu Fabra does not accept any responsibility or liability for data or material
contained on third party sites that reference the information on OpenBMAT Dataset or for the use
any person makes of such third party information. Universitat Pompeu Fabra does not monitor
these third party information and makes no representations in relation to the quality or accuracy
of the information on third party websites or Data Bases.

Use

The data is available for its use and downloadable only for non-profit purposes.

Confidentiality

According to Data Protection Rules only anonymous data are available to users. Each user is
responsible and liable before third Parties to avoid the use of personal data in the OpenBMAT
Dataset, to undertake the necessary measures to ensure anonymity of the data and to foresee
that anonymity is also maintained after merging of data files.