Connectionists: Making Sense of Sounds Data Challenge

[m.plumbley at surrey.ac.uk]

Dear Connectionists,
This data challenge may be of interest to people working on applications
in audio and/or small datasets. Best wishes, Mark Plumbley

-----
 
We hereby announce the "Making Sense of Sounds" (MSoS) Challenge:
 
http://cvssp.org/projects/making_sense_of_sounds/site/challenge/
 
The task in the MSoS Challenge is to classify audio files as
belonging to one of five broad categories derived from human 
classification experiments: Nature, Human, Music, Effects, or Urban.
 
The MSoS Challenge has a development dataset of 1500 five-second
audio files. Performance will be judged using an evaluation dataset of 
500 audio files.
 
The results of the MSoS Challenge will be announced at the DCASE 2018
Workshop:
 
http://dcase.community/workshop2018/
 
For more information about the challenge and how to take part, see:
 
http://cvssp.org/projects/making_sense_of_sounds/site/challenge/
 
Important dates:
 
Challenge announcement and development data set release: 8 Aug 2018 
Evaluation data set release: 1 Oct 2018 
Submission open: 1 Oct 2018 
Submission deadline: 30 Oct 2018 
Results announced: 19/20 Nov 2018 (at DCASE 2018 Workshop)
 
Contact: MSoS.challenge at gmail.com
 
We look forward to your submission!
 
Oliver Bones
On behalf of the MSoS Challenge organizers
 
Additional information:
 
Humans (with no hearing impairment) use sound in everyday life
constantly to interpret their surrounding environment, refocus their
attention, detect anomalies and communicate through language and vocal
emotional expressions. They are able to identify a large number of
sounds, e.g., the call of a bird, the noise of an engine, the cry of a
baby, the sound of a string instrument. They are also capable of
generalising from past experience to new sounds, e.g. recognising a
dulcimer or a kora as a musical instrument despite having never heard
this instrument before in their life. The MSoS data challenge calls
for machine systems to attempt to replicate this human ability.
 
The task is to classify audio data as belonging to one of five broad
categories, which were derived from human classification. In a
psychological experiment at the University of Salford, participants
were asked to categorise 60 sound types, chosen so as to represent the
most commonly used search terms on Freesound.org. Five principal
categories were identified by correspondence analysis and hierarchical
cluster analysis of the human data:
 
Nature
Human
Music
Effects
Urban
 
Within each class the data for the task consists of varying sound
types, e.g., different animals in the 'Nature' category or
different instruments in the 'Music' category such as 'guitar'
and 'mandolin'. Most of the sound types are represented by several
instances themselves, coming from different recordings, e.g. different
guitars. The machine classifier is therefore forced to reproduce a
human capability to be successful: Humans are able to identify a
hitherto unheard animal sound as belonging to an animal based upon
previously established schemas, and a hitherto unheard musical
instrument as a musical instrument, etc.
 
Full details can be found on the MSoS website: 
 
http://cvssp.org/projects/making_sense_of_sounds/site/challenge/

--
Prof Mark D Plumbley
Professor of Signal Processing
Centre for Vision, Speech and Signal Processing (CVSSP) 
University of Surrey, Guildford, Surrey, GU2 7XH, UK
Email: m.plumbley at surrey.ac.uk

=============================================================================
 DCASE 2018 - Detection and Classification of Acoustic Scenes and Events
 Challenge: 30 March - 31 July 2018
 Workshop:  19 - 20 November 2018, Surrey, UK
 http://dcase.community/workshop2018/  http://dcase.community/challenge2018/
=============================================================================