The current neural network based anti-spoofing systems have poor robustness. Their performance degrades further after voice activity detection (VAD) performed, making it difficult to be applied in practice. This work investigated the effect of silence at the beginning and end of speech, finding that silent differences are part of the basis for countermeasures’ judgements. The reason for the performance deterioration caused by VAD is also explored. The experimental results demonstrate that the neural network loses the information about silent segments after the VAD operation removes them. This can lead to more serious overfitting. In order to solve the overfitting problem, the work in this paper also analyzes the reasons for system overfitting from different frequency sub-bands. It is found that the highfrequency part of the feature is the main cause of system overfitting, while the low-frequency part is more robust but less accurate against known attacks. Therefore, we propose the dualband fusion anti-spoofing algorithm, which requires only two sub-systems but outperforms all but one primary system submitted to the logical access condition of the ASVspoof 2019 challenge. Our system has an EER of 3.50% even after VAD operations performed, thus can be put into practical application.

The Effect of Silence and Dual-Band Fusion in Anti-Spoofing System

Spectrograms - time-frequency representations of audio signals - have found widespread use in neural network-based spoofing detection. While deep models are trained on the fullband spectrum of the signal, we argue that not all frequency bands are useful for these tasks. In this paper, we systematically investigate the impact of different subbands and their importance on replay spoofing detection on two benchmark datasets: ASVspoof 2017 v2.0 and ASVspoof 2019 PA. We propose a joint subband modelling framework that employs n different sub-networks to learn subband specific features. These are later combined and passed to a classifier and the whole network weights are updated during training. Our findings on the ASVspoof 2017 dataset suggest that the most discriminative information appears to be in the first and the last 1 kHz frequency bands, and the joint model trained on these two subbands shows the best performance outperforming the baselines by a large margin. However, these findings do not generalise on the ASVspoof 2019 PA dataset. This suggests that the datasets available for training these models do not reflect real world replay conditions suggesting a need for careful design of datasets for training replay spoofing countermeasures.

/pdf/subband-modeling-for-spoofing-detection-in-automatic-speaker-3xz4q0er71.pdf

Subband modeling for spoofing detection in automatic speaker verification

The threat of spoofing can pose a risk to the reliability of automatic speaker verification. Results from the bi-annual ASVspoof evaluations show that effective countermeasures demand front-ends designed specifically for the detection of spoofing artefacts. Given the diversity in spoofing attacks, ensemble methods are particularly effective. The work in this paper shows that a bank of very simple classifiers, each with a front-end tuned to the detection of different spoofing attacks and combined at the score level through non-linear fusion, can deliver superior performance than more sophisticated ensemble solutions that rely upon complex neural network architectures. Our comparatively simple approach outperforms all but 2 of the 48 systems submitted to the logical access condition of the most recent ASVspoof 2019 challenge.

Spoofing Attack Detection using the Non-linear Fusion of Sub-band Classifiers

The automatic Speaker Verification (ASV) system has gained popularity due to its application in the area of biometrics. ASV system’s performance deteriorates because of spoofing attacks. In this paper, types of spoofing attacks are presented. The performance of ASV system depends on the front-end processing which includes extraction of features. This paper presents the contribution of researchers to the detection of various spoofing attacks. The research community has drawn attention to that among the various spoofing attacks that affect the ASV systems, a replay attack is challenging. In this paper, recent approaches for the detection of replay attack are presented.

Countermeasures and Challenges for Detection of Spoofing Attacks in Automatic Speaker Verification System

Automatic speaker verification systems have been widely employed in a variety of commercial applications. However, advancements in the field of speech technology have equipped the attackers with sophisticated techniques for circumventing speaker verification systems. The state-of-the-art countermeasures are fairly successful in detecting speech synthesis and voice conversion attacks. However, the problem of replay attack detection has not received much attention from the researchers. In this study, we perform subband analysis on constant-Q cepstral coefficient (CQCC) and mel-frequency cepstral coefficient (MFCC) features to improve the performance of replay attack detection. We have performed experiments on the ASVspoof 2017 database which consists of 3566 genuine and 15380 replay utterances. Our experimental results suggest that the features extracted from the high frequency band carries significant discriminatory information for replay attack detection. In particular, our approach achieves an improvement of 36.33% over the baseline replay attack detection method in terms of equal error rate.

Sachin Garg

Papers

Subband Analysis for Performance Improvement of Replay Attack Detection in Speaker Verification Systems