[CogSci] ZeroSpeech 2021 challenge, 2nd edition.

[Bertrand HIGY]

Dear colleagues, We are pleased to officially announce the second round of
the 2021 Zero Resource Speech Challenge, on spoken language modelling,
which will take the form of a competition at NeurIPS. This enhanced edition
introduces a new track exploring multimodal language learning, which might
be of interest for members of the Cognitive Science Society. Unlike ASR
systems, human children achieve their language skills without direct
supervision, simply interacting with their physical and linguistic
environments. These interactions are essentially multimodal, with children
leveraging the rich information provided by the context to learn to process
speech. This motivated the new edition of the ZeroSpeech challenge which,
in contrast to the earlier editions that have purely focused on audio-based
learning of linguistic representations, takes a step towards multimodal
language learning by asking participants to train on audiovisual data. We
invite: - Substantive development of existing systems doing spoken language
modelling. - Systems exploring the utility of weak supervision from images
or videos (Track 2). - New approaches to spoken language modeling (with or
without visual supervision) and in particular cognitively inspired methods.
For more detailed information, please see the ZeroSpeech 2021 website:
https://zerospeech.com/2021/ Please get in touch with us if you have
questions or comments about this continuation of the 2021 Zero Resource
Speech Challenge, and we look forward to seeing the results! Best, The Zero
Resource Speech Challenge Organizers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cognitivesciencesociety.org/pipermail/announcements-cognitivesciencesociety.org/attachments/20210707/1092b725/attachment.html>