ASVspoof 2021 (PA)

Authors:
Junichi Yamagishi, Xin Wang, Massimiliano Todisco, Md Sahidullah, Jose Patino, Andreas Nautsch, Xuechen Liu, Kong Aik Lee, Tomi Kinnunen, Nicholas Evans, Hector Delgado

Abstract:
Involves replay attacks recorded in real physical spaces with various noise and reverberation conditions.

Data Creation Method:
Bona fide utterances are made in a real, physical space where spoofing attacks are captured and then replayed within the same physical space using replay devices of varying quality.

Number of Speakers:

• Training and Development: 30 speakers (20 for training, 10 for development)
• Evaluation: 48 speakers (21 male, 27 female)

Total Size:

• Not specified

Number of Real Samples:

• Not specified

Number of Fake Samples:

• More than 100 different spoofing algorithms

Description of the Dataset:

• The dataset includes bona fide and spoofed utterances recorded and replayed in real physical spaces with varying conditions.

Extra Details:
The best performing system achieved a minimum t-DCF of 0.6824 and an EER of 24.25%.

Data Type:

• PCM files

Average Length:

• Not specified

Keywords:

• Speaker verification, Spoofing, Anti-spoofing, Countermeasure, Deepfake detection

When Published:

• 2021

Annotation Process:
Genuine utterances were recorded from speakers in controlled environments. Spoofed utterances were generated using various speech synthesis and voice conversion algorithms.

Usage Scenarios:
Evaluating deepfake detection systems and anti-spoofing countermeasures.

Miscellaneous Information:
The dataset provides a challenging benchmark with various replay attacks in different physical conditions.

Credits:
Datasets Used:

• Replay attacks recorded in real physical spaces

Speech Synthesis Models Referenced:

• Various speech synthesis and voice conversion algorithms