Submission

Preparing your submission

A submission to the challenge consists of
  • either an extended 2-page abstract or a full paper up to 6 pages (in case you addressed one track) or 8 pages (in case you addressed both tracks) describing your system and reporting its performance on the development and test sets
  • the recognition output transcripts on these two sets (.mlf files)

Novelty of the proposed system is not required: established approaches are welcome. Accepted papers will be presented at a dedicated workshop that will be held in conjunction with ICASSP 2013.

Manuscripts should be prepared using the same format as for ICASSP 2013 submissions. Please use the author kit from the ICASSP site, which includes templates for LaTeX and Microsoft Word.

Specific guidelines

Check instructions for Track 1 and Track 2.

In the case when you used other data than provided in the training and development sets, please also report results which involve only those data provided in the training and development sets.

In the case when you used a different language model than provided in the baseline, please also report the results of your system with the provided baseline language model.

In the case when your system can be split into a front end and a back end and the latter differs from the provided baseline back end, you should also report the results of

  • the baseline back-end system trained on denoised training data obtained with your own front end
  • one of the provided pre-trained baseline back-end systems for clean, reverberated or noisy data (the best model among these is enough), only using your own front end to denoise the test data
The interface between front and back end is taken to be either at the signal or feature level, depending whether your front end operates in the signal or feature domain.

Submitting your paper

Please submit your paper to the CHiME workshop following the instructions on the workshop paper submission page.