*** apologies for any cross-postings ***
Dear colleagues,
We’re excited to announce the release of soundata,
a python library for reproducible use of audio datasets.
Soundata can be installed via: pip install soundata
The source code lives here: https://github.com/soundata/soundata
We’re launching with 14
popular environmental sound datasets, with plans to continue expanding with additional datasets spanning a range of audio domains, including bioacoustics!
Soundata makes it easy to:
-
Download datasets to a common location and format
-
Validate that a downloaded dataset is complete and perfectly matches a canonical version
-
Load audio and annotation files into a common format
-
Parse clip-level metadata for detailed evaluations
We hope soundata will help the community to:
-
Ensure results are reproducible by working against exactly the
same data
-
Save time by avoiding manual downloads and having to write custom dataset parsers
-
Automate large-scale download, training, and evaluation pipelines
-
Increase the visibility of new datasets by adding them to soundata
Soundata is a cross-organizational collaboration spanning researchers from , Adobe
Research, ,
and .
You can learn more about the library on our docs page: https://soundata.readthedocs.io/
A bit more about the motivation for soundata can be found in our (work in progress) paper:
"Soundata: A Python library for reproducible use of audio datasets"
Magdalena Fuentes, Justin Salamon, Pablo Zinemanas, Martín Rocamora, Genís Plaja, Irán R. Román, Marius Miron, Xavier Serra, Juan Pablo Bello
[arXiv]
We *welcome and encourage* contributions from the community, especially data loaders for datasets not included yet in soundata. If you'd be interested in adding a bioacoustics dataset
to soundata, we'd love to hear from you!
Cheers,
Justin & Magdalena on behalf of the soundata team