Blogs

Freesound Audio Tagging 2019 Challenge

(Original text at the web of the Music Technology Group)

The Freesound team of the MTG, in collaboration with the Sound Understanding team at Google AI launch the Freesound Audio Tagging 2019 Challenge in Kaggle

Some sounds are distinct and instantly recognizable, like a baby’s laugh or the strum of a guitar. Other sounds are difficult to pinpoint. If you close your eyes, could you tell the difference between the sound of a chainsaw and the sound of a blender?

Because of the vastness of sounds we experience, no reliable automatic general-purpose audio tagging systems exist. A significant amount of manual effort goes into tasks like annotating sound collections and providing captions for non-speech events in audiovisual content.

To tackle this problem, Freesound and Google AI's Sound Understanding team have teamed up to develop the dataset for this new competition: the Freesound Audio Tagging 2019 Challenge.

This is a follow-up of last year's successful Freesound General-Purpose Audio Tagging Challenge. In this new edition, we are taking the challenge to the next level with multi-label audio tagging, twice the number of audio categories, and a noisier-than-ever training set. Participants are challenged to build systems able to recognize 80 diverse categories of everyday sounds. A small subset of FSD is used as a part of the data in the competition.

The competition is hosted on Kaggle (a well-known platform that hosts machine learning competitions) and under the framework of the DCASE Challenge (an academic challenge featuring tasks related to the computational analysis of sound events).

Find more information and join the competition here: https://www.kaggle.com/c/freesound-audio-tagging-2019/