No abstract

This paper describes our DKU replay detection system for the ASVspoof 2019 challenge. The goal is to develop spoofing countermeasure for automatic speaker recognition in physical access scenario. We leverage the countermeasure system pipeline from four aspects, including the data augmentation, feature representation, classification, and fusion. First, we introduce an utterance-level deep learning framework for antispoofing. It receives the variable-length feature sequence and outputs the utterance-level scores directly. Based on the framework, we try out various kinds of input feature representations extracted from either the magnitude spectrum or phase spectrum. Besides, we also perform the data augmentation strategy by applying the speed perturbation on the raw waveform. Our best single system employs a residual neural network trained by the speed-perturbed group delay gram. It achieves EER of 1.04% on the development set, as well as EER of 1.08% on the evaluation set. Finally, using the simple average score from several single systems can further improve the performance. EER of 0.24% on the development set and 0.66% on the evaluation set is obtained for our primary system.

The ongoing ASVspoof 2017 challenge aims to detect replay attacks for text dependent speaker verification. In this paper, we propose multiple replay spoofing countermeasure systems, with some of them boosting the CQCC-GMM baseline system after score level fusion. We investigate different steps in the system building pipeline, including data augmentation, feature representation, classification and fusion. First, in order to augment training data and simulate the unseen replay conditions, we converted the raw genuine training data into replay spoofing data with parametric sound reverberator and phase shifter. Second, we employed the original spectrogram rather than C-QCC as input to explore the end-to-end feature representation learning methods. The spectrogram is randomly cropped into fixed size segments, and then fed into a deep residual netowrk (ResNet). Third, upon the CQCC features, we replaced the subsequent GMM classifier with deep neural networks including fully-connected deep neural network (FDNN) and Bidirectional Long Short Term Memory neural network (BLST-M). Experiments showed that data augmentation strategy can significantly improve the system performance. The final fused system achieves to 16.39 % EER on the test set of ASVspoof 2017 for the common task.

In this paper, we propose an iterative framework for self-supervised speaker representation learning based on a deep neural network (DNN). The framework starts with training a self-supervision speaker embedding network by maximizing agreement between different segments within an utterance via a contrastive loss. Taking advantage of DNN's ability to learn from data with label noise, we propose to cluster the speaker embedding obtained from the previous speaker network and use the subsequent class assignments as pseudo labels to train a new DNN. Moreover, we iteratively train the speaker network with pseudo labels generated from the previous step to bootstrap the discriminative power of a DNN. Speaker verification experiments are conducted on the VoxCeleb dataset. The results show that our proposed iterative self-supervised learning framework outperformed previous works using self-supervision. The speaker network after 5 iterations obtains a 61% performance gain over the speaker embedding model trained with contrastive loss.

Despite the significant improvements in speaker recognition enabled by deep neural networks, unsatisfactory performance persists under noisy environments. In this paper, we train the speaker embedding network to learn the "clean" embedding of the noisy utterance. Specifically, the network is trained with the original speaker identification loss with an auxiliary within-sample variability-invariant loss. This auxiliary variability-invariant loss is used to learn the same embedding among the clean utterance and its noisy copies and prevents the network from encoding the undesired noises or variabilities into the speaker representation. Furthermore, we investigate the data preparation strategy for generating clean and noisy utterance pairs on-the-fly. The strategy generates different noisy copies for the same clean utterance at each training step, helping the speaker embedding network generalize better under noisy environments. Experiments on VoxCeleb1 indicate that the proposed training framework improves the performance of the speaker verification system in both clean and noisy conditions.