ASSERT: Anti-Spoofing with Squeeze-Excitation and Residual neTworks

Lai, Cheng-I; Chen, Nanxin; Villalba, Jesús; Dehak, Najim

doi:10.48550/arxiv.1904.01120

Cited by 15 publications

(18 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…System #1 refers to the proposed architecture that jointly optimizes SID, PAD, and ISV loss, Figure 1-(a). System #2-SE is the result of applying squeezeexcitation (SE) [26] based on its recent application to PAD [9]. System #3 describes the result of assigning three max feature map (MFM) blocks [21] for SID as well as for PAD after the first three MFM blocks.…”

Section: Results Analysismentioning

confidence: 99%

“…However, SV systems are known to be vulnerable to spoofing attacks such as replay attacks, voice conversion, and speech synthesis. These vulnerabilities have inspired research into presentation attack detection (PAD), which classifies given utterances as spoofed or not spoofed [6][7][8]; notably, many DNN-based systems have achieved promising results [9][10][11].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Integrated Replay Spoofing-aware Text-independent Speaker Verification

Shim,

Jung,

Kim

et al. 2020

Preprint

View full text Add to dashboard Cite

A number of studies have successfully developed speaker verification or spoofing detection systems. However, studies integrating the two tasks remain in the preliminary stages. In this paper, we propose two approaches for the integrated replay spoofingaware speaker verification task: an end-to-end monolithic and a back-end modular approach. The first approach simultaneously trains speaker identification, replay spoofing detection, and the integrated system using multi-task learning with a common feature. However, through experiments, we hypothesize that the information required for performing speaker verification and replay spoofing detection might differ because speaker verification systems try to remove device-specific information from speaker embeddings while replay spoofing exploits such information. Therefore, we propose a back-end approach using a deep neural network that takes speaker embeddings extracted from enrollment and test utterances and a replay detection prediction on the test utterance as input. Experiments are conducted using the ASVspoof 2017-v2 dataset, which includes official trials on the integration of speaker verification and replay spoofing detection. The proposed back-end approach demonstrates a relative improvement of 21.77% in terms of the equal error rate for integrated trials compared to a conventional speaker verification system.

show abstract

Section: Results Analysismentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Integrated Replay Spoofing-aware Text-independent Speaker Verification

Shim,

Jung,

Kim

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…We follow the pre-training settings as in [66], where we pre-train our Mockingjay model on 360 hours of speech on the LibriSpeech dataset [72]. Two highperformance anti-spoofing models are adopted: LCNN [55] and SENet [56]. The implementation details of the two models can be found in [62].…”

Section: Experiments Setupmentioning

confidence: 99%

“…Spoofing countermeasure models, also known as anti-spoofing models, are shields for ASV to detect and filter spoofing audio. Recently several high performance antispoofing models have been proposed [47][48][49][50][51][52][53][54][55][56].…”

Section: Introductionmentioning

confidence: 99%

Defense for Black-box Attacks on Anti-spoofing Models by Self-Supervised Learning

Liu

Lee

2020

Preprint

View full text Add to dashboard Cite

High-performance anti-spoofing models for automatic speaker verification (ASV), have been widely used to protect ASV by identifying and filtering spoofing audio that is deliberately generated by text-to-speech, voice conversion, audio replay, etc. However, it has been shown that high-performance antispoofing models are vulnerable to adversarial attacks. Adversarial attacks, that are indistinguishable from original data but result in the incorrect predictions, are dangerous for antispoofing models and not in dispute we should detect them at any cost. To explore this issue, we proposed to employ Mockingjay, a self-supervised learning based model, to protect antispoofing models against adversarial attacks in the black-box scenario. Self-supervised learning models are effective in improving downstream task performance like phone classification or ASR. However, their effect in defense for adversarial attacks has not been explored yet. In this work, we explore the robustness of self-supervised learned high-level representations by using them in the defense against adversarial attacks. A layerwise noise to signal ratio (LNSR) is proposed to quantize and measure the effectiveness of deep models in countering adversarial noise. Experimental results on the ASVspoof 2019 dataset demonstrate that high-level representations extracted by Mockingjay can prevent the transferability of adversarial examples, and successfully counter black-box attacks.

show abstract

“…Some methods also use raw signals to extract features using methods such as SincNet [18] or Variational Auto Encoder (VAE) [19]. The second category deals with a variety of classifiers such as Neural networkbased methods including VGG [18], Squeeze-Excitation (SE), Residual network, Siamese networks [20], and recurrent networks [21], as well as other traditional GMM-based methods. Certain methods have also used end-to-end structures for this purpose.…”

Section: Introductionmentioning

confidence: 99%

Efficient Attention Branch Network with Combined Loss Function for Automatic Speaker Verification Spoof Detection

Rostami¹,

Homayounpour²,

Nickabadi³

2021

Preprint

View full text Add to dashboard Cite

Many endeavors have sought to develop countermeasure techniques as enhancements on Automatic Speaker Verification (ASV) systems, in order to make them more robust against spoof attacks. As evidenced by the latest ASVspoof 2019 countermeasure challenge, models currently deployed for the task of ASV are, at their best, devoid of suitable degrees of generalization to unseen attacks. Upon further investigation of the proposed methods, it appears that a broader three-tiered view of the proposed systems; comprised of the classifier, feature extraction phase, and model loss function, may to some extent lessen the problem. Accordingly, the present study proposes the Efficient Attention Branch Network (EABN) modular architecture with a combined loss function to address the generalization problem. The EABN architecture is based on attention and perception branches; the purpose of the attention branch-also interpretable from a human's point of view-is to produce an attention mask meant to improve classification performance. The perception branch, on the other hand, is used for the primary purpose of the problem at hand, that is, spoof detection. The new EfficientNet-A0 architecture was employed for the perception branch, with nearly ten times fewer parameters and approximately seven times fewer floating-point operations than the top performing SE-Res2Net50 network. The final evaluation results on ASVspoof 2019 dataset suggest an EER = 0.86% and t-DCF = 0.0239 in the Physical Access (PA) scenario using the log-PowSpec input feature, the EfficientNet-A0 for the perception branch, and the combined loss function. Furthermore, using the LFCC input feature, the SE-Res2Net50 for the perception branch, and the combined loss function, the proposed model performed at figures of EER = 1.89% and t-DCF = 0.507 in the Logical Access (LA) scenario, which to the best of our knowledge, is the best single system ASV spoofing countermeasure.

show abstract

ASSERT: Anti-Spoofing with Squeeze-Excitation and Residual neTworks

Cited by 15 publications

References 27 publications

Integrated Replay Spoofing-aware Text-independent Speaker Verification

Integrated Replay Spoofing-aware Text-independent Speaker Verification

Defense for Black-box Attacks on Anti-spoofing Models by Self-Supervised Learning

Efficient Attention Branch Network with Combined Loss Function for Automatic Speaker Verification Spoof Detection

Contact Info

Product

Resources

About