fsdnoisy18k

Created by Eduardo Fonseca, Mercedes Collado, Manoj Plakal, Daniel P. W. Ellis, Frederic Font, Xavier Favory, and Xavier Serra

version	1.0.0
license	CC-BY-3.0
usage	unrestricted
languages
format	wav
channel	1
sampling rate	44100
bit depth	16
duration	1 days 18:29:50.962131518
files	18532, duration distribution: 0.3 s 30.0 s
repository	audb-public

Description

FSDnoisy18k is an audio dataset collected with the aim of fostering the investigation of label noise in sound event classification. It contains 42.5 hours of audio across 20 sound classes, including a small amount of manually-labeled data and a larger quantity of real-world noisy data. FSDnoisy18k contains 18,532 audio clips (42.5h) unequally distributed in the 20 classes drawn from the AudioSet Ontology. The audio clips are of variable length ranging from 300ms to 30s, and each clip has a single ground truth label (singly-labeled data). Files are released under either CC-BY or CC0. The individual license for each clip is available from the files table. FSDnoisy18k is an expandable dataset that features a per-class varying degree of types and amount of label noise. The dataset allows investigation of label noise as well as other approaches, from semi-supervised learning, e.g., self-training to learning with minimal supervision.

Example

audio/train/432015.wav

Tables

Click on a row to toggle a preview.

ID	Type	Columns
files	filewise	license	file license audio/test/274679.wav CC0-1.0 audio/test/365220.wav CC0-1.0 audio/test/233458.wav CC-BY-3.0 audio/test/370931.wav CC-BY-3.0 audio/test/137172.wav CC-BY-3.0 18532 rows x 1 column
test	filewise	label	file label audio/test/274679.wav Walk, footsteps audio/test/365220.wav Walk, footsteps audio/test/233458.wav Walk, footsteps audio/test/370931.wav Walk, footsteps audio/test/137172.wav Walk, footsteps 947 rows x 1 column
train	filewise	label, manually_verified, noisy_small	file label manually_verified noisy_small audio/train/94322.wav Walk, footsteps True False audio/train/85602.wav Walk, footsteps True False audio/train/240356.wav Walk, footsteps True False audio/train/371015.wav Walk, footsteps True False audio/train/264589.wav Walk, footsteps True False 17585 rows x 3 columns

Schemes

ID	Dtype	Labels	Mappings
categories	str	Acoustic guitar, Bass guitar, Clapping, Coin (dropping), Crash cymbal, Dishes, pots, and pans, Engine, […], Piano, Rain, Slam, Squeak, Tearing, Walk, footsteps, Wind, Writing	child_ids, citation_uri, description, id, positive_examples, restrictions
license	str	CC-BY-3.0, CC0-1.0	✓
manually_verified	bool
noisy_small	bool