ASVspoof 5: Evaluation of Spoofing, Deepfake, and Adversarial Attack Detection Using Crowdsourced Speech

TL;DR

ASVspoof 5 evaluates speech spoofing, deepfake, and adversarial attack detection using a new crowdsourced database with diverse speakers and conditions, analyzing challenge results and future directions.

Contribution

Introduces a new crowdsourced database for ASVspoof 5 and provides comprehensive analysis of detection solutions and challenges.

Findings

Many solutions perform well on standard data

Performance degrades under adversarial attacks

Neural encoding/compression schemes reduce detection accuracy

Abstract

ASVspoof 5 is the fifth edition in a series of challenges which promote the study of speech spoofing and deepfake detection solutions. A significant change from previous challenge editions is a new crowdsourced database collected from a substantially greater number of speakers under diverse recording conditions, and a mix of cutting-edge and legacy generative speech technology. With the new database described elsewhere, we provide in this paper an overview of the ASVspoof 5 challenge results for the submissions of 53 participating teams. While many solutions perform well, performance degrades under adversarial attacks and the application of neural encoding/compression schemes. Together with a review of post-challenge results, we also report a study of calibration in addition to other principal challenges and outline a road-map for the future of ASVspoof.