News and Updates

• 01.09.2020 - challenge results: https://arxiv.org/pdf/2109.00648.pdf and https://hal.archives-ouvertes.fr/hal-03335126v5
• 10.11.2020 - add ASV and ASR models trained on anonymized data for the primary baseline (asr_eval_b1_anon and asv_eval_b1_anon in models.tar.gz) https://www.voiceprivacychallenge.org/vp2020/docs/VoicePrivacy2020_post_evaluation.pdf
• 30.08.2020 - add ZEBRA (Zero Evidence Biometric Recognition Assessment: https://gitlab.eurecom.fr/nautsch/zebra) metrics: expected empirical cross-entropy (on the population and individual level). Paper: https://arxiv.org/pdf/2005.09413.pdf
• 29.08.2020 - add De-Identification & Gain of voice distinctiveness metrics. Paper: https://arxiv.org/pdf/2008.13144.pdf
• 30.07.2020 - add linkability metric.
• 13.05.2020 - updated RESULTS_baseline (plda) and added RESULTS_baseline_cosine (cosine)
• 13.05.2020 - version 1.3 of the evaluation plan: https://www.voiceprivacychallenge.org/vp2020/docs/VoicePrivacy_2020_Eval_Plan_v1_3.pdf
• 17.03.2020 - version 1.2 of the evaluation plan: https://www.voiceprivacychallenge.org/vp2020/docs/VoicePrivacy_2020_Eval_Plan_v1_2.pdf
• 16.03.2020 - clean-up script to delete data in order to rerun anonymization and evaluation: cleanup.sh
• 15.03.2020 - bugfix and updated results for Baseline-1
• 05.03.2020 - script to convert the results file to the latex format results_to_latex.py for the Interspeech paper
• 05.03.2020 - release of a new additional Baseline-2: anonymization using McAdams coefficient
• 17.02.2020 - updated RESULTS with results on development and test data
• 16.02.2020 - baseline is adapted to process test data
• 15.02.2020 - test data release