Published April 5, 2022 | Version 1.0
Dataset Open

USM Dataset - A Dataset for Polyphonic Sound Event Tagging in Urban Sound Monitoring Scenarios

Authors/Creators

  • 1. Fraunhofer IDMT

Description

This dataset includes 24,000 5-seconds-long polyphonic stereo soundscapes composed of sounds taken from the FSD50k dataset:

- Eduardo Fonseca, Xavier Favory, Jordi Pons, Frederic Font, Xavier Serra. FSD50K: an Open Dataset of Human-Labeled Sound Events (https://arxiv.org/abs/2010.00475)

FSD50k samples used in the USM dataset were selected to allow for commercial usage.

Find more details about the USM dataset at https://github.com/jakobabesser/USM

Files

usm_eval.zip

Files (47.7 GB)

Name Size Download all
md5:d9f54a0b78c8bda2df583adc166704d3
4.2 GB Preview Download
md5:5c6f2a5a04ea203ba8cc04b191c89b7c
5.4 GB Download
md5:b598f8a598f9f614c0ce71f7265d442e
5.4 GB Download
md5:d65456e2e75c75b462fedfce5f2b5a64
5.4 GB Download
md5:0e631eaeb8a42f359f6d1417a976e080
5.4 GB Download
md5:6e9c6b6c8930d04ae4b94cbe07179ce9
5.4 GB Download
md5:1bc2cbebc502d163e03b547c15d13259
5.4 GB Download
md5:49ce06fba1437480903b84742a67e937
5.4 GB Download
md5:42738a16beccdca7b4e9bbb469fbaf2f
2.0 GB Preview Download
md5:bd3c6a7c568bb2dc965062e059f5b807
3.8 GB Preview Download

Additional details

References

  • Jakob Abeßer: Classifying Sounds in Polyphonic Urban Sound Scenes, Proceedings of the 152nd AES Convention (2022).