RedDots Challenge

The RedDots Challenge: Towards Characterizing Speakers from Short Utterances

The RedDots project was initiated, with collaboration from multiple sites, as a 
follow‐up to a special session during INTERSPEECH 2014. It was set out to 
collect speech data through mobile crowd‐sourcing, with the benefit of 
potentially wider population and greater diversity. The project was rolled out in 
January 29, 2015. At the time of writing, the project has recruited 89 speakers 
(72 male, 17 female) from 21 countries, with a total of 875 complete sessions. 
The purpose of this special session is to gather the research efforts towards a 
common goal of exploring new directions and better understanding of speaker-channel‐
phonetic variability modeling for text dependent and text‐prompted 
speaker verification over short utterances. Papers submitted to the special 
session are encouraged to provide results based on the RedDots database in 
order to enable comparison of algorithms and methods. For this purpose, the 
organizers strongly encourage the participants to report performance on the 
protocol delivered with the database in terms of EER and minimum cost (in the 
sense of NIST 2008 and/or 2010 Speaker Recognition Evaluation). To get the 
database, please contact the organizers or visit the pages as listed below.

List of potential topics

The special session topics include but not limited to:
  • Speaker characterization and adaptation for short duration
  • Text‐dependent speaker characterization
  • Content matching for long‐enrolment short‐test speaker verification
  • Robustness in channels and environment for short duration segments
  • End‐to‐end speaker‐channel‐text modelling using DNN/RNN
  • Evaluation of text‐dependent speaker verification
  • Features for text‐dependent speaker verification
  • Robustness to template aging

Important dates

  • 30 Dec 2015 reddots_r2015q4_v1 dataset release
  • 23 Mar 2016 Interspeech 2016 paper submission deadline

Important links

  • RedDots Project:
  • RedDots dataset:
  • Discussion group: text‐dependent‐


Call for Participation
  • The call for participation in PDF format can be downloaded below.
  • [Jan 13, 2016] Speaker Info List available
  • [Jan 27, 2016] Utterance transcription available (for registered users only)
  • [Feb 06, 2016] Baseline for Part04 TD and TP tasks available
  • [Feb 06, 2016] Updated readme available
Kong Aik Lee,
Jan 12, 2016, 5:12 PM