Adversarial Network Bottleneck Features for Noise Robust Speaker Verification

Abstract

In this paper, we propose a noise robust bottleneck feature representation which is generated by an adversarial network (AN). The AN includes two cascade connected networks, an encoding network (EN) and a discriminative network (DN). Melfrequency cepstral coefficients (MFCCs) of clean and noisy speech are used as input to the EN and the output of the EN is used as the noise robust feature. The EN and DN are trained in turn, namely, when training the DN, noise types are selected as the training labels and when training the EN, all labels are set as the same, i.e., the clean speech label, which aims to make the AN features invariant to noise and thus achieve noise robustness. We evaluate the performance of the proposed feature on a Gaussian Mixture Model-Universal Background Model based speaker verification system, and make comparison to MFCC features of speech enhanced by short-time spectral amplitude minimum mean square error (STSA-MMSE) and deep neural network-based speech enhancement (DNN-SE) methods. Experimental results on the RSR2015 database show that the proposed AN bottleneck feature (AN-BN) dramatically outperforms the STSA-MMSE and DNN-SE based MFCCs for different noise types and signal-to-noise ratios. Furthermore, the AN-BN feature is able to improve the speaker verification performance under the clean condition.

Time-Contrastive Learning Based Deep Bottleneck Features for Text-Dependent Speaker Verification

“…To obtain the final BN feature, the output from a hidden layer, a 1024 dimensional deep feature, is projected onto a 57 dimensional space to align with the dimension of the MFCC feature for a fair comparison. Allowing a higher dimension for BN can potentially boost the performance as observed in [4]. Deep features are normalized to zero mean and unit variance at utterance level before using principle component analysis (PCA) for dimension reduction.…”

A Survey on Machine Learning‐Based Mobile Big Data Analysis: Challenges and Applications

“…Recently, some adversarial training methods are introduced to extract noise invariant bottleneck features [64,188]. As shown in Figure 12, the adversarial network includes two parts, i.e., an encoding network (EN) which can extract noise invariant features and a discriminative network (DN) which can judge noise types of the noise invariant feature generated from EN.…”

“…As shown in Figure 12, the adversarial network includes two parts, i.e., an encoding network (EN) which can extract noise invariant features and a discriminative network (DN) which can judge noise types of the noise invariant feature generated from EN. Therefore, we can get robustness noise invariant features from EN which can improve the performance of speaker verification system by adversarial training these two parts in turn [64,188].…”

“…The transmission data focus on solving channel modeling [55,56] and user access problems corresponding to the physical transmission system of M-Internet. On this foundation, application data focus on the applications based on the MBD including social networks analysis [57][58][59], user behavior analysis [48,50,60], speech analysis and decision in IoV [61][62][63][64][65][66], smart grid [67,68], networked healthcare [53,69,70], finance services [46,71], etc.…”

Cross-lingual Text-independent Speaker Verification Using Unsupervised Adversarial Discriminative Domain Adaptation

“…Recently, the speaker representation models have moved from the commonly used i-vector model [1,2,3], with a probabilistic linear discriminant (PLDA) back-end [4,5] to a new paradigm: speaker embedding trained from deep neural networks. Various speaker embeddings based on different network architectures [6,7] , attention mechanism [8,9], loss functions [10,11], noise robustness [12,13], and training paradigms [14,15] have been proposed and greatly improve the performance of speaker verification systems. Snyder et al [6] recently proposed the x-vector model, which is based on a Time-Delay Deep Neural Network (TDNN) architecture that computes speaker embeddings from variable-length acoustic segments.…”