The ASVspoof 2017 Challenge: Assessing the Limits of Replay Spoofing Attack Detection

Abstract

The ASVspoof initiative was created to promote the development of countermeasures which aim to protect automatic speaker verification (ASV) from spoofing attacks. The first community-led, common evaluation held in 2015 focused on countermeasures for speech synthesis and voice conversion spoofing attacks. Arguably, however, it is replay attacks which pose the greatest threat. Such attacks involve the replay of recordings collected from enrolled speakers in order to provoke false alarms and can be mounted with greater ease using everyday consumer devices. ASVspoof 2017, the second in the series, hence focused on the development of replay attack countermeasures. This paper describes the database, protocols and initial findings. The evaluation entailed highly heterogeneous acoustic recording and replay conditions which increased the equal error rate (EER) of a baseline ASV system from 1.76% to 31.46%. Submissions were received from 49 research teams, 20 of which improved upon a baseline replay spoofing detector EER of 24.77%, in terms of replay/non-replay discrimination. While largely successful, the evaluation indicates that the quest for countermeasures which are resilient in the face of variable replay attacks remains very much alive.

Speech Replay Spoofing Attack Detection System Based on Fusion of Classification Algorithms

“…In this study, we carried out our experiments using ASV Spoof 2017 corpus [1], which is provided as a part of spoofing challenge. It originates from the RedDots corpus [3], which was collected by volunteers from across the globe.…”

“…The primary performance measure is the equal error rate (EER) [1,4]. We obtain the scores from our classification algorithms.…”

“…It has to meet requirements of robustness to changes in acoustic environment or changes in voice of target speakers. However, fast development of modern technologies of voice generation, digital processing and recording leads to the fact that it is necessary to take into account the potential threats from the so-called voice spoofing attacks [1,2]. The first and most probable of the attack vectors is a replay attack.…”

“…It has to distinguish acoustic conditions in genuine and recorded voice, noise properties of the acoustic channel, and possible changes in voice spectral characteristics. The promotion of the development of effective replay attack countermeasures was the main motivation for organizing the ASV Spoof Challenge 2017 [1]. This challenge focuses on a standalone replay attack detection.…”

“…We opted for some common speaker modelling techniques such as Gaussian mixture models (GMM) [5], GMM based universal background models (UBM) and i-vectors [6,7]. The set of classifiers consists of multiple GMMs based classifier [1,5], i-vector-Gaussian Probabilistic Linear Discriminant Analysis (GPLDA) [7] and highly popular in machine learning community tree boosting method XGBoost [8]. The most common choice of an acoustic feature for voice processing system is Mel frequency cepstral coefficients (MFCCs) [9].…”