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

[AUDITORY] SiSEC 2018: Community-Based Signal Separation Evaluation Campaign

                    *** CALL FOR PARTICIPATION ***

                   Community-Based Signal Separation
                          Evaluation Campaign
                             (SiSEC 2018)

                        https://sisec.inria.fr/

   Deadline for submission of separation results: April, 1st 2018

Following the success of the six previous editions, we are glad to
announce the launch of the SiSEC 2018 campaign for the evaluation of
source separation algorithms. This campaign is open to anyone, whether you
are an expert or a newcomer, and whether your algorithm relies on prior
information or not.

This is a community-based endeavor: all aspects of evaluation, including
datasets and evaluation criteria, are agreed between the participants.

*A SPECIAL SISEC SESSION WILL TAKE PLACE AT LVA/ICA 2018*
participants are encouraged to present their methods there.
When submitting your paper at LVA/ICA 2018, pick
"Special Session: SiSEC 2018".
See more information at http://cvssp.org/events/lva-ica-2018/


This year, SiSEC features two audio tasks:

1) Professionally produced music recordings (MUS), Following the success
of the previous editions, we worked a lot to improve the music separation
task. This year's edition includes:

- An augmented dataset, with 100 full songs for training, 50 songs for
  test, all mixed professionally in stereo, for anyone to use.

- This year, all the data is encoded in the Native Instruments STEMS
  format, yielding only 4.4gb for 150 tracks, while maintaining trans-
  parent quality.

- A set of Python tools to load, parse, and use the dataset easily. An
  easy-to-use standalone utility is also provided to convert the data to
  WAV if needed.

- Baseline scripts implementing oracle methods, including Ideal Binary
  Mask, Ideal Ratio Mask, and Multichannel Wiener Filter.

- Submission will be automated, and analysis plots will be generated
  automatically.

2) Asynchronous recordings of speech mixtures (ASY), runs with the same
datasets and metrics as SiSEC 2015.


Please visit https://sisec.inria.fr/ for further instructions.

IMPORTANT DATES

Jan 15th 2018 - LVA/ICA 2018 Paper submission deadline
Mar 1st 2018  - Submission system for SiSEC open
Apr 1st 2018  - Submission of separation results
Apr 8th 2018  - Publication of separation results
Apr 16 2018   - LVA/ICA Camera-ready papers deadline
Jun 2nd 2018  - LVA/ICA 2018 conference begins.

Please register to the mailing list sisec[at]inria.fr at
https://sympa.inria.fr/sympa/info/sisec

If you have any question, do not hesitate to contact us.

Best regards,

SiSEC 2018 Chair
Antoine Liutkus

PS: We are sorry in case of cross posting.