[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

4th CHiME Speech Separation and Recognition Challenge and Workshop

       4th CHiME Speech Separation and Recognition
                 Challenge and Workshop

                 Launch: April 27, 2016
                Deadline: August 19, 2016

      Workshop: Google, San Francisco, Sep 13, 2016


Dear colleague,

It gives us great pleasure to announce the official launch of the CHiME-4 Challenge and Workshop.

CHiME-4 revisits the CHiME-3 data, i.e., Wall Street Journal sentences spoken by talkers in challenging noisy environments, with updated baselines and fewer microphones for test. Participants may submit to one or more tracks: 1-channel, 2-channel, and 6-channel.


The Challenge website is now live and contains all the information, tools and data that you will need for participation:
- a detailed description of the challenge scenario and recording conditions,
- real training and development data,
- baseline software for data simulation, speech enhancement and state-of-the-art Kaldi-based speech recognition,
- full instructions for participation and submission,
- answers to frequently asked questions.

If you have any question that isn't answered by the website please email chimechallenge@xxxxxxxxx

We look forward to your participation.


If you are considering participating, please email chimechallenge@xxxxxxxxx with subject 'CHiME4' and you will be added to the email list for receiving further updates.


27th April, 2016  —  Launch - Training/dev data and baselines released
10th June, 2016 — Workshop registration open (limited seats available per institution, preferably for challenge participants)
27th June, 2016   —  Test data released
19th Aug, 2016    —  Challenge abstract/paper submission deadline
24th Aug, 2016    —  Paper notification
13th Sept, 2016   —  CHiME-4 Workshop
14th Oct, 2016    —  Final paper (2 to 6 pages)


Emmanuel Vincent, Inria
Shinji Watanabe, MERL
Jon Barker & Ricard Marxer, University of Sheffield


Kean Chin, Google


Mitsubishi Electric Research Laboratories (MERL)