General-purpose Tagging of Freesound Audio with AudioSet Labels: Task Description, Dataset, and Baseline

By Eduardo Fonseca, Manoj Plakal, Frederic Font, D. Ellis, Xavier Favory, Jordi Pons, Xavier Serra. 2018

167 citations recorded by the source corpus.

This paper describes Task 2 of the DCASE 2018 Challenge, titled "General-purpose audio tagging of Freesound content with AudioSet labels". This task was hosted on the Kaggle platform as "Freesound General-Purpose Audio Tagging Challenge". The goal of the task is to build an audio tagging system that can recognize the category of an audio clip from a subset of 41 diverse categories drawn from the AudioSet Ontology. We present the task, the dataset prepared for the competition, and a baseline system.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Signal Processing

Citations in the corpus

FSD50K: An Open Dataset of Human-Labeled Sound Events (2020, methodology): “specifically, DCASE 2018 Task 2 [55], DCASE 2019 Tasks”
PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition (2019, methodology): “The mAP@3 is used for evaluating system performance [54].”
Learning Sound Event Classifiers from Web Audio with Noisy Labels (2019, background): “In a recent audio tagging Kaggle competition using FSDKaggle2018 [5], a number of approaches were proposed to deal with the label noise present.”
Convolutional Recurrent Neural Network and Data Augmentation for Audio Tagging with Noisy Labels and Minimal Supervision (2019, background): “The DCASE 2019 Challenge Task 2 “Audio Tagging with noisy labels and minimal supervision” [10] is a follow-up of the DCASE 2018 Challenge Task 2 “General-purpose audio tagging of Freesound content with AudioSet labels” [11].”
Unsupervised Contrastive Learning of Sound Event Representations (2020, unknown): “We use the FSDnoisy18k dataset [21] 23, 24, 25, 21, 26]).”
Model-Agnostic Approaches To Handling Noisy Labels When Training Sound Event Classifiers (2019, unknown): “, FSDKaggle2018 [4] or FSDnoisy18k [5]) to thousands (e.”
Training general-purpose audio tagging networks with noisy labels and iterative self-verification (2018, background): “Available: https://arxiv.org/abs/1807.09840 [5] H. Eghbal-Zadeh, B. Lehner, M. Dorfer, and G. Widmer, “Cpjku submissions for dcase-2016: A hybrid approach using binaural i-vectors and deep convolutional neural networks,” IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Even”
Audio tagging with noisy labels and minimal supervision (2019, methodology): “This generalizes the mean reciprocal rank used in the 2018 version of the challenge [3] to the case of multiple true labels per 1 https://www.kaggle.com/c/freesound-audio-tagging-2019 Note that the competition name on Kaggle is abbreviated from the full DCASE2019 Challenge task name to “Freesound…”
Facilitating the Manual Annotation of Sounds When Using Large Taxonomies (2018, unknown): “Instead of focusing on the recognition of a speciﬁc subset of sounds, general-purpose taxonomies enable tasks that aim to recognise and describe a wider (and usually more generic) range of sounds [10].”
An Adversarial Feature Distillation Method for Audio Classification (2019, background): “The DCASE18 task 2 [34] is for general-purpose audio tagging task.”
Addressing Missing Labels in Large-Scale Sound Event Recognition Using a Teacher-Student Framework With Loss Masking (2020, background): “As Sound Event Recognition (SER) has gained increasing attention in recent years [1], research in this field has moved from small datasets encompassing few hours of audio [2, 3, 4, 5], to larger datasets with much greater coverage and duration [6, 7].”