ASVspoof 2019

Automatic Speaker Verification

Spoofing and Countermeasures Challenge

Future horizons in spoofed/fake audio detection


17th July: Release of ASVspoof 2019 real PA database. In this public release, we extended the simulated data used in the challenge with a real small set of audio files (2700) recorded and replayed in 3 different labs. To download it, please visit our Licence page.

5th June: Database Release via Edinburgh DataShare.

2nd May: ASRU 2019 Special Session description.

7th Feb: read the ASVspoof 2019 press alert.

15th Jan: the ASVspoof 2019 evaluation plan v0.4 is now available (ChangeLog).
 

The ASVspoof 2019 registration procedure is described in the evaluation plan. Registrations will be confirmed by email within 48 hours. The registration confirmation email will also contain a link and credentials for the downloading of training and development data.

The ASVspoof 2019 schedule is as follows.

19th December, 2018 Training and development data
8th February, 2019 Participant registration deadline
15th February, 2019 Evaluation data
22nd February, 2019 Scores submission
15th March, 2019 Results
29th March, 2019 Interspeech submission deadline
15th-19th September, 2019 ASVspoof 2019 special session at Interspeech 2019

ASVspoof 2019 participants may wish to read our Odyssey 2018 paper which describes the t-DCF metric. It will be the default metric for ASVspoof 2019.

T. Kinnunen, K.-A. Lee, H. Delgado, N. Evans, M. Todisco, M. Sahidullah, J. Yamagishi, D.-A. Reynolds, "t-DCF: a Detection Cost Function for the Tandem Assessment of Spoofing Countermeasures and Automatic Speaker Verification", Proc. Odyssey 2018 - The Speaker and Language Recognition Workshop [PDF]

Software implementations of the t-DCF metric can be downloaded in MATLAB and Python formats.


Participants are encouraged to subscribe to the ASVspoof 2019 mailing list.

Send an email to sympa@asvspoof.org with 'subscribe asvspoof2019' in the subject line
 

ASVspoof 2019 participants may make use of the following baseline countermeasures:

M. Sahidullah, T. Kinnunen and C. Hanilci, "A comparison of features for synthetic speech detection", Proc. INTERSPEECH 2015, pp. 2087–2091 [PDF]

M. Todisco, H. Delgado and N. Evans, "Constant Q cepstral coefficients: a spoofing countermeasure for automatic speaker verification", Computer, Speech and Language, vol. 45, pp. 516 –535, 2017 [PDF]

Software implementations of both the LFCC-GMM and CQCC-GMM baselines are also available.

 

Organisers (equal contribution)

Junichi Yamagishi, National Institute of Informatics, JAPAN / University of Edinburgh, UK
Massimiliano Todisco, EURECOM, FRANCE
Md Sahidullah, Inria, FRANCE
Héctor Delgado, EURECOM, FRANCE
Xin Wang, National Institute of Informatics, JAPAN
Nicholas Evans, EURECOM, FRANCE
Tomi Kinnunen, University of Eastern Finland, FINLAND
Kong Aik Lee, NEC, JAPAN
Ville Vestman, University of Eastern Finland, FINLAND
Andreas Nautsch, EURECOM, France

 
EURECOM Inria National Institute of Informatics
NEC University of Eastern Finland The University of Edinburgh

Contributors to database creation

 
Aalto University Academia Sinica DFKI DFKI EURECOM
Google Google Google News Initiative HOYA iFlaytek
Nagoya University National Institute of Informatics Nippon Telegraph and Telephone National Institute of Informatics
Universität des Saarlandes University of Eastern Finland Université d'Avignon The University of Edinburgh USTC
 

Funding agencies

 
Academy of Finland Agence Nationale de la Recherche Japan Science and Technology Agency