ASVspoof 2019: A large-scale public database of synthesized, converted and replayed speech

By Xin Eric Wang, Junichi Yamagishi, Massimiliano Todisco, Héctor Delgado, Andreas Nautsch, N. Evans, Md. Sahidullah, Ville Vestman, Tomi H. Kinnunen, Kong Aik LEE and 9 more. Computer Speech and Language, 2019

651 citations recorded by the source corpus.

Abstract Automatic speaker verification (ASV) is one of the most natural and convenient means of biometric person recognition. Unfortunately, just like all other biometric systems, ASV is vulnerable to spoofing, also referred to as “presentation attacks.” These vulnerabilities are generally unacceptable and call for spoofing countermeasures or “presentation attack detection” systems. In addition to impersonation, ASV systems are vulnerable to replay, speech synthesis, and voice conversion attacks. The ASVspoof challenge initiative was created to foster research on anti-spoofing and to provide common platforms for the assessment and comparison of spoofing countermeasures. The first edition, ASVspoof 2015, focused upon the study of countermeasures for detecting of text-to-speech synthesis (TTS) and voice conversion (VC) attacks. The second edition, ASVspoof 2017, focused instead upon replay spoofing attacks and countermeasures. The ASVspoof 2019 edition is the first to consider all three spoofing attack types within a single challenge. While they originate from the same source database and same underlying protocol, they are explored in two specific use case scenarios. Spoofing attacks within a logical access (LA) scenario are generated with the latest speech synthesis and voice conversion technologies, including state-of-the-art neural acoustic and waveform model techniques. Replay spoofing attacks within a physical access (PA) scenario are generated through carefully controlled simulations that support much more revealing analysis than possible previously. Also new to the 2019 edition is the use of the tandem detection cost function metric, which reflects the impact of spoofing and countermeasures on the reliability of a fixed ASV system. This paper describes the database design, protocol, spoofing attack implementations, and baseline ASV and countermeasure results. It also describes a human assessment on spoofed data in logical access. It was demonstrated that the spoofing data in the ASVspoof 2019 database have varied degrees of perceived quality and similarity to the target speakers, including spoofed data that cannot be differentiated from bona fide utterances even by human subjects. It is expected that the ASVspoof 2019 database, with its varied coverage of different types of spoofing data, could further foster research on anti-spoofing.

Subfields: Artificial Intelligence, Signal Processing

View Paper: https://www.semanticscholar.org/paper/0f79266f1c9f6fe0d0f8abcca138d19073c013da

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentation (2022, positive): “We used the training and development partitions of the ASVspoof 2019 LA database [61, 62] for training and validation.”
ASVspoof 2021: Towards Spoofed and Deepfake Speech Detection in the Wild (2022, methodology): “1, [18]) which is already difficult to detect even in the absence of encoding, becomes disproportionately more difficult to detect after low bandwidth encoding.”
End-to-End anti-spoofing with RawNet2 (2020, methodology): “Spoofed attacks are generated using a set of 19 different speech synthesis, voice conversion and hybrid algorithms [23].”
Creating, Using, Misusing, and Detecting Deep Fakes (2022, methodology): “Wang et al.’s forensic classifier (S.-Y. Wang et al. 2020), for example, is based on a standard neural-network architecture (ResNet-50), pre-trained on a standard image database (ImageNet), and then refined to classify an image as real or synthesized.”
An Overview of Voice Conversion and Its Challenges: From Statistical Modeling to Deep Learning (2020, methodology): “The best performing systems in the 2016 and 2018 voice conversion challenges were also used for generating advanced spoofed audio [302].”
FakeAVCeleb: A Novel Audio-Video Multimodal Deepfake Dataset (2021, methodology): “Google proposed the Automatic Speaker Verification Spoofing (ASV) [41] challenge dataset with the goal of speaker verification and spoofed voice detection.”
Translatotron 2: High-quality direct speech-to-speech translation with voice preservation (2021, background): “ASVspoof 2019 (Todisco et al., 2019; Wang et al., 2020) found that it was challenging to detect spoof audios generated from a zero-shot voice cloning TTS model (Jia et al.”
One-Class Learning Towards Synthetic Voice Spoofing Detection (2020, background): “The ASVspoof 2019 challenge provides a standard database [30] for anti-spoofing.”
AASIST: Audio Anti-Spoofing Using Integrated Spectro-Temporal Graph Attention Networks (2021, methodology): “All experiments were performed using the ASVspoof 2019 logical access (LA) dataset [3, 39].”
Translatotron 2: Robust direct speech-to-speech translation (2021, methodology): “Todisco et al. (2019); Wang et al. (2020b) showed that it was challenging to detect spoof audios generated from Jia et al. (2018), which uses the same speaker encoder-based approach as in the original Translatotron.”
ASVspoof 2021: accelerating progress in spoofed and deepfake speech detection (2021, methodology): “With no new training or development data being released, participants were expected to use the training and development partitions of the ASVspoof 2019 databases [9] based upon speech derived from the VCTK base corpus(4).”