urbansound8k

Created by Justin Salamon, Christopher Jacoby, Juan Pablo Bello

version	1.0.0
license	CC-BY-NC-3.0
usage	research
languages
format	wav
channel	1, 2
sampling rate	8000, 11024, 11025, 16000, 22050, 24000, 32000, 44100, 48000, 96000, 192000
bit depth	4, 8, 16, 24, 32
duration	0 days 08:45:00.880612979
files	8732, duration distribution: 0.1 s 4.0 s
repository	audb-public

Description

The UrbanSound8k dataset contains 8732 labeled sound excerpts (<=4s) of urban sounds from 10 classes. All excerpts are taken from field recordings uploaded to Freesound. The files are pre-sorted into ten folds. The sampling rate, bit depth, and number of channels are the same as those of the original file uploaded to Freesound and may vary from file to file.

Example

audio/fold6/101281-3-0-5.wav

Tables

Click on a row to toggle a preview.

ID	Type	Columns
files	filewise	category, clip_id, fold, salience	file category clip_id fold salience audio/fold5/100032-3-0-0.wav dog_bark 100032 5 1 audio/fold5/100263-2-0-117.wav children_playing 100263 5 1 audio/fold5/100263-2-0-121.wav children_playing 100263 5 1 audio/fold5/100263-2-0-126.wav children_playing 100263 5 1 audio/fold5/100263-2-0-137.wav children_playing 100263 5 1 8732 rows x 4 columns

Schemes

ID	Dtype	Labels	Mappings
category	str	air_conditioner, car_horn, children_playing, dog_bark, drilling, engine_idling, gun_shot, jackhammer, siren, street_music
clip_id	int
fold	int	1, 2, 3, 4, 5, 6, 7, 8, 9, 10
salience	int	1, 2	✓