The First VoicePrivacy Attacker Challenge

Training, development, and evaluation datasets were provided along with a baseline attacker . Participants developed their attacker systems in the form of automatic speaker verification systems and submitted their scores on the development and evaluation data. The best attacker systems reduced the equal error rate (EER) by 25-44% relative w.r .t. the baseline. Index T erms --V oice privacy, voice anonymization, attacker system, automatic speaker verification I. C ONTEXT Speech conveys a lot of personal data, e.g., age and gender, health, geographical or ethnic origin, and socio-economic status. Formed in 2020, the V oicePrivacy initiative [1] promotes privacy enhancing solutions for speech technology via a series of benchmarking challenges.