End-to-end spectro-temporal graph attention networks for speaker verification anti-spoofing and speech deepfake detection
Сквозные спектрально-временные графовые сети с механизмом внимания для защиты верификации говорящего от спуфинга и обнаружения дипфейков речи
2021-09-10
SCID: 54.1/s25m28f2
Discuss with AI
end-to-end neural networksspeaker verification anti-spoofingspectro-temporal graph attention networksspeech deepfake detection
Figures from the paper
Abstract (AI)
Artefacts that serve to distinguish bona fide speech from spoofed or deepfake speech are known to reside in specific subbands and temporal segments. Various approaches can be used to capture and model such artefacts, however, none works well across a spectrum of diverse spoofing attacks. Reliable detection then often depends upon the fusion of multiple detection systems, each tuned to detect different forms of attack. In this paper we show that better performance can be achieved when the fusion is performed within the model itself and when the representation is learned automatically from raw waveform inputs. The principal contribution is a spectro-temporal graph attention network (GAT) which learns the relationship between cues spanning different sub-bands and temporal intervals. Using a model-level graph fusion of spectral (S) and temporal (T) sub-graphs and a graph pooling strategy to improve discrimination, the proposed RawGAT-ST model achieves an equal error rate of 1.06 % for the ASVspoof 2019 logical access database. This is one of the best results reported to date and is reproducible using an open source implementation.
Key Findings
1
The abstract provides no quantitative results, benchmark comparisons, or explicitly stated limitations.
2
The paper presents an end-to-end spectro-temporal graph attention network for speaker-verification anti-spoofing and speech-deepfake detection.
3
The proposed approach jointly models spectro-temporal relationships in speech using graph attention mechanisms.
Research Object
speaker verification anti-spoofing and speech deepfake detection systems
Research Subject
end-to-end spectro-temporal graph attention modeling for distinguishing bona fide speech from spoofed and deepfake speech
Publication Details
Publication Date
2021-09-10
Journal
Publisher
ISSN
Cited by
255
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai6
Deep Residual Learning for Image Recognition2016
Adam: A Method for Stochastic Optimization2014
Semi-Supervised Classification with Graph Convolutional Networks2016
A Comprehensive Survey on Graph Neural Networks2020
The Graph Neural Network Model2008
RawNet: Advanced End-to-End Deep Neural Network Using Raw Waveforms for Text-Independent Speaker Verification2019