An Audio Fingerprinting Approach to Replay Attack Detection on ASVSPOOF 2017 Challenge Data

Abstract

Replay attacks, where an impostor replays a genuine user utterance, are a major vulnerability of speaker verification systems. Two highly likely scenarios for replay attacks are either hidden recording of actual spoken access trials, or reusing previous genuine recordings in case of fraudulent access to transmission channels or storage devices. In both scenarios, an audio fingerprint-based approach comparing any access trial with all previous recordings from the claimed speaker perfectly fits the task of replay attack detection. However, ASVspoof 2017 rules did not allow the use of the original RedDots audio files (spoofed trials are replayed versions of RedDots), which disabled a fingerprint-based regular participation in the evaluation as those original files are necessary to build the bank of previous-access audio fingerprints. Then, we agreed with the organizers to run and submit on time a parallel fingerprintbased evaluation with exactly the same blind test data with an alternative but realistic (deployable) evaluation scenario. While we obtained an Equal Error Rate of 8.91% detecting replayed versus genuine trials, this result is not comparable for ranking purposes with those from actual participants in the Challenge as we used the original RedDots files. However, it provides insight into the potential and complementarity of audio fingerprinting, especially for high audio-quality attacks where state-of-the-art acoustic antispoofing systems show poor performance (the best ASVspoof 2017 system with global EER of 6.73% degraded to about 25% in condition C6 of high-quality replays), while our fingerprint-based antispoofer obtains an EER of 0.0% for the high-quality replays in condition C6, showing the complementarity of acoustic antispoofers for low-mid quality replays and fingerprint-based ones for mid-high quality replays.

Audio deepfakes: A survey

“…Some of the replay attack detection systems have been proposed by working on the features which are fed into the network (Witkowski et al, 2017 ). Others have improved the networks used or have worked on both of the networks and features (Lavrentyeva et al, 2017 ; Nagarsheth et al, 2017 ; Gonzalez-Rodriguez et al, 2018 ; Huang and Pun, 2019 , 2020 ; Lai et al, 2019 ; Li et al, 2019 ). Additionally, before the ASVspoof Challenge 2017 (Kinnunen et al, 2017 ; Lavrentyeva et al, 2017 ), there were only a couple of research papers done on replay attack, and after this challenge, more approaches for this attack were researched (Tom et al, 2018 ; Pradhan et al, 2019 ).…”

Detecting Spoofing Attacks Using VGG and SincNet: BUT-Omilia Submission to ASVspoof 2019 Challenge

“…The last type of attacks is replay attack with pre-recorded audio and it is considered to be the most difficult attack to detect [1]. Possible ways to tackle this problem are (a) anti-spoofing techniques based on detecting typical distortions in recorded and replayed audio [3,9], (b) using audio fingerprinting [10] to detect a replay of an enrollment utterance, and (c) using liveness detection and phrase verification [11] in text-dependent speaker verification. This paper presents the collaborative efforts of BUT and Omilia to introduce novel countermeasures for the last three attack types, as part of the 2019 automatic speaker verification (ASV) anti-spoofing challenge.…”

Transforming acoustic characteristics to deceive playback spoofing countermeasures of speaker verification systems

“…The third type utilizes audio fingerprinting to check whether an incoming recording is similar to previously authenticated utterances that were automatically saved in the ASV system. Rodriguez et al [13] developed such a system: if the similarity score was higher than a threshold, the recording was treated as a playback attack. A disadvantage of this type of CM is that it is sensitive to noise.…”