Showing papers on "Speaker recognition published in 2021"

ASVspoof 2019: Spoofing Countermeasures for the Detection of Synthesized, Converted and Replayed Speech

Abstract: The ASVspoof initiative was conceived to spearhead research in anti-spoofing for automatic speaker verification (ASV). This paper describes the third in a series of bi-annual challenges: ASVspoof 2019. With the challenge database and protocols being described elsewhere, the focus of this paper is on results and the top performing single and ensemble system submissions from 62 teams, all of which out-perform the two baseline systems, often by a substantial margin. Deeper analyses shows that performance is dominated by specific conditions involving either specific spoofing attacks or specific acoustic environments. While fusion is shown to be particularly effective for the logical access scenario involving speech synthesis and voice conversion attacks, participants largely struggled to apply fusion successfully for the physical access scenario involving simulated replay attacks. This is likely the result of a lack of system complementarity, while oracle fusion experiments show clear potential to improve performance. Furthermore, while results for simulated data are promising, experiments with real replay data show a substantial gap, most likely due to the presence of additive noise in the latter. This finding, among others, leads to a number of ideas for further research and directions for future editions of the ASVspoof challenge.

The Idlab Voxsrc-20 Submission: Large Margin Fine-Tuning and Quality-Aware Score Calibration in DNN Based Speaker Verification

Abstract: In this paper we propose and analyse a large margin fine-tuning strategy and a quality-aware score calibration in text-independent speaker verification. Large margin fine-tuning is a secondary training stage for DNN based speaker verification systems trained with margin-based loss functions. It enables the network to create more robust speaker embeddings by enabling the use of longer training utterances in combination with a more aggressive margin penalty. Score calibration is a common practice in speaker verification systems to map output scores to well-calibrated log-likelihood-ratios, which can be converted to interpretable probabilities. By including quality features in the calibration system, the decision thresholds of the evaluation metrics become quality-dependent and more consistent across varying trial conditions. Applying both enhancements on the ECAPA-TDNN architecture leads to state-of-the-art results on all publicly available VoxCeleb1 test sets and contributed to our winning submissions in the supervised verification tracks of the Vox-Celeb Speaker Recognition Challenge 2020.

FluentNet: End-to-End Detection of Stuttered Speech Disfluencies With Deep Learning

Abstract: Millions of people are affected by stuttering and other speech disfluencies, with the majority of the world having experienced mild stutters while communicating under stressful conditions. While there has been much research in the field of automatic speech recognition and language models, stutter detection and recognition has not received as much attention. To this end, we propose an end-to-end deep neural network, FluentNet, capable of detecting a number of different stutter types. FluentNet consists of a Squeeze-and-Excitation Residual convolutional neural network which facilitate the learning of strong spectral frame-level representations, followed by a set of bidirectional long short-term memory layers that aid in learning effective temporal relationships. Lastly, FluentNet uses an attention mechanism to focus on the important parts of speech to obtain a better performance. We perform a number of different experiments, comparisons, and ablation studies to evaluate our model. Our model achieves state-of-the-art results by outperforming other solutions in the field on the publicly available UCLASS dataset. Additionally, we present LibriStutter: a stuttered speech dataset based on the public LibriSpeech dataset with synthesized stutters. We also evaluate FluentNet on this dataset, showing the strong performance of our model versus a number of baseline and state-of-the-art techniques.

Who is Real Bob? Adversarial Attacks on Speaker Recognition Systems

Abstract: Speaker recognition (SR) is widely used in our daily life as a biometric authentication or identification mechanism. The popularity of SR brings in serious security concerns, as demonstrated by recent adversarial attacks. However, the impacts of such threats in the practical black-box setting are still open, since current attacks consider the white-box setting only.In this paper, we conduct the first comprehensive and systematic study of the adversarial attacks on SR systems (SRSs) to understand their security weakness in the practical black-box setting. For this purpose, we propose an adversarial attack, named FAKEBOB, to craft adversarial samples. Specifically, we formulate the adversarial sample generation as an optimization problem, incorporated with the confidence of adversarial samples and maximal distortion to balance between the strength and imperceptibility of adversarial voices. One key contribution is to propose a novel algorithm to estimate the score threshold, a feature in SRSs, and use it in the optimization problem to solve the optimization problem. We demonstrate that FAKEBOB achieves 99% targeted attack success rate on both open-source and commercial systems. We further demonstrate that FAKEBOB is also effective on both open-source and commercial systems when playing over the air in the physical world. Moreover, we have conducted a human study which reveals that it is hard for human to differentiate the speakers of the original and adversarial voices. Last but not least, we show that four promising defense methods for adversarial attack from the speech recognition domain become ineffective on SRSs against FAKEBOB, which calls for more effective defense methods. We highlight that our study peeks into the security implications of adversarial attacks on SRSs, and realistically fosters to improve the security robustness of SRSs.

A Deep Neural Network Model for Speaker Identification

Abstract: Speaker identification is a classification task which aims to identify a subject from a given time-series sequential data. Since the speech signal is a continuous one-dimensional time series, most of the current research methods are based on convolutional neural network (CNN) or recurrent neural network (RNN). Indeed, these methods perform well in many tasks, but there is no attempt to combine these two network models to study the speaker identification task. Due to the spectrogram that a speech signal contains, the spatial features of voiceprint (which corresponds to the voice spectrum) and CNN are effective for spatial feature extraction (which corresponds to modeling spectral correlations in acoustic features). At the same time, the speech signal is in a time series, and deep RNN can better represent long utterances than shallow networks. Considering the advantage of gated recurrent unit (GRU) (compared with traditional RNN) in the segmentation of sequence data, we decide to use stacked GRU layers in our model for frame-level feature extraction. In this paper, we propose a deep neural network (DNN) model based on a two-dimensional convolutional neural network (2-D CNN) and gated recurrent unit (GRU) for speaker identification. In the network model design, the convolutional layer is used for voiceprint feature extraction and reduces dimensionality in both the time and frequency domains, allowing for faster GRU layer computation. In addition, the stacked GRU recurrent network layers can learn a speaker’s acoustic features. During this research, we tried to use various neural network structures, including 2-D CNN, deep RNN, and deep LSTM. The above network models were evaluated on the Aishell-1 speech dataset. The experimental results showed that our proposed DNN model, which we call deep GRU, achieved a high recognition accuracy of 98.96%. At the same time, the results also demonstrate the effectiveness of the proposed deep GRU network model versus other models for speaker identification. Through further optimization, this method could be applied to other research similar to the study of speaker identification.

A Survey of Speaker Recognition: Fundamental Theories, Recognition Methods and Opportunities

Abstract: Humans can identify a speaker by listening to their voice, over the telephone, or on any digital devices. Acquiring this congenital human competency, authentication technologies based on voice biometrics, such as automatic speaker recognition (ASR), have been introduced. An ASR recognizes speakers by analyzing speech signals and characteristics extracted from speaker’s voices. ASR has recently become an effective research area as an essential aspect of voice biometrics. Specifically, this literature survey gives a concise introduction to ASR and provides an overview of the general architectures dealing with speaker recognition technologies, and upholds the past, present, and future research trends in this area. This paper briefly describes all the main aspects of ASR, such as speaker identification, verification, diarization etc. Further, the performance of current speaker recognition systems are investigated in this survey with the limitations and possible ways of improvement. Finally, a few unsolved challenges of speaker recognition are presented at the closure of this survey.

On Joint Optimization of Automatic Speaker Verification and Anti-Spoofing in the Embedding Space

Abstract: Biometric systems are exposed to spoofing attacks which may compromise their security, and voice biometrics based on automatic speaker verification (ASV), is no exception. To increase the robustness against such attacks, anti-spoofing systems have been proposed for the detection of replay, synthesis and voice conversion-based attacks. However, most proposed anti-spoofing techniques are loosely integrated with the ASV system. In this work, we develop a new integration neural network which jointly processes the embeddings extracted from ASV and anti-spoofing systems in order to detect both zero-effort impostors and spoofing attacks. Moreover, we propose a new loss function based on the minimization of the area under the expected (AUE) performance and spoofability curve (EPSC), which allows us to optimize the integration neural network on the desired operating range in which the biometric system is expected to work. To evaluate our proposals, experiments were carried out on the recent ASVspoof 2019 corpus, including both logical access (LA) and physical access (PA) scenarios. The experimental results show that our proposal clearly outperforms some well-known techniques based on the integration at the score- and embedding-level. Specifically, our proposal achieves up to 23.62% and 22.03% relative equal error rate (EER) improvement over the best performing baseline in the LA and PA scenarios, respectively, as well as relative gains of 27.62% and 29.15% on the AUE metric.

Microsoft Speaker Diarization System for the Voxceleb Speaker Recognition Challenge 2020

Abstract: This paper describes the Microsoft speaker diarization system for monaural multi-talker recordings in the wild, evaluated at the diarization track of the VoxCeleb Speaker Recognition Challenge (VoxSRC) 2020. We will first explain our system design to address issues in handling real multi-talker recordings. We then present the details of the components, which include Res2Net-based speaker embedding extractor, conformer-based continuous speech separation with leakage filtering, and a modified DOVER (short for Diarization Output Voting Error Reduction) method for system fusion. We evaluate the systems with the data set provided by VoxSRC challenge 2020, which contains real-life multi-talker audio collected from YouTube. Our best system achieves 3.71% and 6.23% of the diarization error rate (DER) on development set and evaluation set, respectively, being ranked the 1st at the diarization track of the challenge.

Black-box Adversarial Attacks on Commercial Speech Platforms with Minimal Information

Abstract: Adversarial attacks against commercial black-box speech platforms, including cloud speech APIs and voice control devices, have received little attention until recent years. The current "black-box" attacks all heavily rely on the knowledge of prediction/confidence scores to craft effective adversarial examples, which can be intuitively defended by service providers without returning these messages. In this paper, we propose two novel adversarial attacks in more practical and rigorous scenarios. For commercial cloud speech APIs, we propose Occam, a decision-only black-box adversarial attack, where only final decisions are available to the adversary. In Occam, we formulate the decision-only AE generation as a discontinuous large-scale global optimization problem, and solve it by adaptively decomposing this complicated problem into a set of sub-problems and cooperatively optimizing each one. Our Occam is a one-size-fits-all approach, which achieves 100% success rates of attacks with an average SNR of 14.23dB, on a wide range of popular speech and speaker recognition APIs, including Google, Alibaba, Microsoft, Tencent, iFlytek, and Jingdong, outperforming the state-of-the-art black-box attacks. For commercial voice control devices, we propose NI-Occam, the first non-interactive physical adversarial attack, where the adversary does not need to query the oracle and has no access to its internal information and training data. We combine adversarial attacks with model inversion attacks, and thus generate the physically-effective audio AEs with high transferability without any interaction with target devices. Our experimental results show that NI-Occam can successfully fool Apple Siri, Microsoft Cortana, Google Assistant, iFlytek and Amazon Echo with an average SRoA of 52% and SNR of 9.65dB, shedding light on non-interactive physical attacks against voice control devices.

ASV-SUBTOOLS: Open Source Toolkit for Automatic Speaker Verification

Abstract: In this paper, we introduce a new open source toolkit for automatic speaker verification (ASV), named ASV-Subtools. Adopting PyTorch as main deep learning engine and Kaldi toolkit for data processing, ASV-Subtools allows users to develop modern speaker recognizers flexibly and efficiently. The toolkit prioritizes efficiency, modularity, and extensibility with the goal of supporting the state-of-the-art technologies in speaker recognition. In addition to including the commonly used networks, such as the time delay neural networks (TDNN), factorized TDNN (F-TDNN) and ResNet, ASV-Subtools also integrates an upgraded version of SpecAugment data augmentation method, named Inverted SpecAugment, with focus on making it more appropriate for speaker recognition subtasks. Besides, for alleviating the domain mismatch between training and test data, ASV-Subtools provides multiple domain adaptation methods of Probabilistic Linear Discriminant Analysis (PLDA). Experimental results show that state-of-the-art techniques implemented on ASV-Subtools could achieve competitive performance compared to other implementations.

Black-box Adversarial Attacks on Commercial Speech Platforms with Minimal Information

Abstract: Adversarial attacks against commercial black-box speech platforms, including cloud speech APIs and voice control devices, have received little attention until recent years. Constructing such attacks is difficult mainly due to the unique characteristics of time-domain speech signals and the much more complex architecture of acoustic systems. The current "black-box" attacks all heavily rely on the knowledge of prediction/confidence scores or other probability information to craft effective adversarial examples (AEs), which can be intuitively defended by service providers without returning these messages. In this paper, we take one more step forward and propose two novel adversarial attacks in more practical and rigorous scenarios. For commercial cloud speech APIs, we propose Occam, a decision-only black-box adversarial attack, where only final decisions are available to the adversary. In Occam, we formulate the decision-only AE generation as a discontinuous large-scale global optimization problem, and solve it by adaptively decomposing this complicated problem into a set of sub-problems and cooperatively optimizing each one. Our Occam is a one-size-fits-all approach, which achieves 100% success rates of attacks (SRoA) with an average SNR of 14.23dB, on a wide range of popular speech and speaker recognition APIs, including Google, Alibaba, Microsoft, Tencent, iFlytek, and Jingdong, outperforming the state-of-the-art black-box attacks. For commercial voice control devices, we propose NI-Occam, the first non-interactive physical adversarial attack, where the adversary does not need to query the oracle and has no access to its internal information and training data. We, for the first time, combine adversarial attacks with model inversion attacks, and thus generate the physically-effective audio AEs with high transferability without any interaction with target devices. Our experimental results show that NI-Occam can successfully fool Apple Siri, Microsoft Cortana, Google Assistant, iFlytek and Amazon Echo with an average SRoA of 52% and SNR of 9.65dB, shedding light on non-interactive physical attacks against voice control devices.

Block-Based High Performance CNN Architectures for Frame-Level Overlapping Speech Detection

Abstract: Speech technology systems such as Automatic Speech Recognition (ASR), speaker diarization, speaker recognition, and speech synthesis have advanced significantly by the emergence of deep learning techniques. However, none of these voice-enabled systems perform well in natural environmental circumstances, specifically in situations where one or more potential interfering talkers are involved. Therefore, overlapping speech detection has become an important front-end triage step for speech technology applications. This is crucial for large-scale datsets where manual labeling in not possible. A block-based CNN architecture is proposed to address modeling overlapping speech in audio streams with frames as short as 25 ms. The proposed architecture is robust to both: (i) shifts in distribution of network activations due to the change in network parameters during training, (ii) local variations from the input features caused by feature extraction, environmental noise, or room interference. We also investigate the effect of alternate input features including spectral magnitude, MFCC, MFB, and pyknogram on both computational time and classification performance. Evaluation is performed on simulated overlapping speech signals based on the GRID corpus. The experimental results highlight the capability of the proposed system in detecting overlapping speech frames with 90.5% accuracy, 93.5% precision, 92.7% recall, and 92.8% Fscore on same gender overlapped speech. For opposite gender cases, the network scores exceed 95% in all the classification metrics.

SoK: The Faults in our ASRs: An Overview of Attacks against Automatic Speech Recognition and Speaker Identification Systems

Abstract: Speech and speaker recognition systems are employed in a variety of applications, from personal assistants to telephony surveillance and biometric authentication. The wide deployment of these systems has been made possible by the improved accuracy in neural networks. Like other systems based on neural networks, recent research has demonstrated that speech and speaker recognition systems are vulnerable to attacks using manipulated inputs. However, as we demonstrate in this paper, the end-to-end architecture of speech and speaker systems and the nature of their inputs make attacks and defenses against them substantially different than those in the image space. We demonstrate this first by systematizing existing research in this space and providing a taxonomy through which the community can evaluate future work. We then demonstrate experimentally that attacks against these models almost universally fail to transfer. In so doing, we argue that substantial additional work is required to provide adequate mitigations in this space.

Hear "No Evil", See "Kenansville"*: Efficient and Transferable Black-Box Attacks on Speech Recognition and Voice Identification Systems

Abstract: Automatic speech recognition and voice identification systems are being deployed in a wide array of applications, from providing control mechanisms to devices lacking traditional interfaces, to the automatic transcription of conversations and authentication of users. Many of these applications have significant security and privacy considerations. We develop attacks that force mistranscription and misidentification in state of the art systems, with minimal impact on human comprehension. Processing pipelines for modern systems are comprised of signal preprocessing and feature extraction steps, whose output is fed to a machine-learned model. Prior work has focused on the models, using white-box knowledge to tailor model-specific attacks. We focus on the pipeline stages before the models, which (unlike the models) are quite similar across systems. As such, our attacks are black-box, transferable, can be tuned to require zero queries to the target, and demonstrably achieve mistranscription and misidentification rates as high as 100% by modifying only a few frames of audio. We perform a study via Amazon Mechanical Turk demonstrating that there is no statistically significant difference between human perception of regular and perturbed audio. Our findings suggest that models may learn aspects of speech that are generally not perceived by human subjects, but that are crucial for model accuracy.

The ins and outs of speaker recognition: lessons from VoxSRC 2020

Abstract: The VoxCeleb Speaker Recognition Challenge (VoxSRC) at Interspeech 2020 offers a challenging evaluation for speaker recognition systems, which includes celebrities playing different parts in movies. The goal of this work is robust speaker recognition of utterances recorded in these challenging environments. We utilise variants of the popular ResNet architecture for speaker recognition and perform extensive experiments using a range of loss functions and training parameters. To this end, we optimise an efficient training framework that allows powerful models to be trained with limited time and resources. Our trained models demonstrate improvements over most existing works with lighter models and a simple pipeline. The paper shares the lessons learned from our participation in the challenge.

UniSpeech-SAT: Universal Speech Representation Learning with Speaker Aware Pre-Training.

Abstract: Self-supervised learning (SSL) is a long-standing goal for speech processing, since it utilizes large-scale unlabeled data and avoids extensive human labeling. Recent years witness great successes in applying self-supervised learning in speech recognition, while limited exploration was attempted in applying SSL for modeling speaker characteristics. In this paper, we aim to improve the existing SSL framework for speaker representation learning. Two methods are introduced for enhancing the unsupervised speaker information extraction. First, we apply the multi-task learning to the current SSL framework, where we integrate the utterance-wise contrastive loss with the SSL objective function. Second, for better speaker discrimination, we propose an utterance mixing strategy for data augmentation, where additional overlapped utterances are created unsupervisely and incorporate during training. We integrate the proposed methods into the HuBERT framework. Experiment results on SUPERB benchmark show that the proposed system achieves state-of-the-art performance in universal representation learning, especially for speaker identification oriented tasks. An ablation study is performed verifying the efficacy of each proposed method. Finally, we scale up training dataset to 94 thousand hours public audio data and achieve further performance improvement in all SUPERB tasks.

X-Vectors: New Quantitative Biomarkers for Early Parkinson's Disease Detection From Speech.

Abstract: Many articles have used voice analysis to detect Parkinson's disease (PD), but few have focused on the early stages of the disease and the gender effect. In this article, we have adapted the latest speaker recognition system, called x-vectors, in order to detect PD at an early stage using voice analysis. X-vectors are embeddings extracted from Deep Neural Networks (DNNs), which provide robust speaker representations and improve speaker recognition when large amounts of training data are used. Our goal was to assess whether, in the context of early PD detection, this technique would outperform the more standard classifier MFCC-GMM (Mel-Frequency Cepstral Coefficients-Gaussian Mixture Model) and, if so, under which conditions. We recorded 221 French speakers (recently diagnosed PD subjects and healthy controls) with a high-quality microphone and via the telephone network. Men and women were analyzed separately in order to have more precise models and to assess a possible gender effect. Several experimental and methodological aspects were tested in order to analyze their impacts on classification performance. We assessed the impact of the audio segment durations, data augmentation, type of dataset used for the neural network training, kind of speech tasks, and back-end analyses. X-vectors technique provided better classification performances than MFCC-GMM for the text-independent tasks, and seemed to be particularly suited for the early detection of PD in women (7-15% improvement). This result was observed for both recording types (high-quality microphone and telephone).

Enhancement in speaker recognition for optimized speech features using GMM, SVM and 1-D CNN

Abstract: Contemporary automatic speaker recognition (ASR) systems do not provide 100% accuracy making it imperative to explore different techniques to improve it. Easy access to mobile devices and advances in sensor technology, has made voice a preferred parameter for biometrics. Here, a comparative analysis of accuracies obtained in ASR with employment of classical Gaussian mixture model (GMM), support vector machine (SVM) which is the machine learning algorithm and the state of art 1-D CNN as classifiers is presented. Authors propose considering dynamic voice features along with static features as relevant speaker information in them lead to substantial improvement in the accuracy for ASR. As concatenation of features leads to the redundancy and increased computation complexity, Fisher score algorithm was employed to select the best contributing features resulting in improvement in accuracy. The results indicate that SVM and 1-D Neural network outperform GMM. Support Vector Machine (SVM), and 1-D CNN gave comparable results with 1-D CNN giving an improved accuracy of 94.77% in ASR.

DeepMSRF: A Novel Deep Multimodal Speaker Recognition Framework with Feature Selection

Abstract: For recognizing speakers in video streams, significant research studies have been made to obtain a rich machine learning model by extracting high-level speaker’s features such as facial expression, emotion, and gender. However, generating such a model is not feasible by using only single modality feature extractors that exploit either audio signals or image frames, extracted from video streams. In this paper, we address this problem from a different perspective and propose an unprecedented multimodality data fusion framework called DeepMSRF, Deep Multimodal Speaker Recognition with Feature selection. We execute DeepMSRF by feeding features of the two modalities, namely speakers’ audios and face images. DeepMSRF uses a two-stream VGGNET to train on both modalities to reach a comprehensive model capable of accurately recognizing the speaker’s identity. We apply DeepMSRF on a subset of VoxCeleb2 dataset with its metadata merged with VGGFace2 dataset. The goal of DeepMSRF is to identify the gender of the speaker first, and further to recognize his or her name for any given video stream. The experimental results illustrate that DeepMSRF outperforms single modality speaker recognition methods with at least 3% accuracy.

Improved Speech Emotion Recognition using Transfer Learning and Spectrogram Augmentation

Abstract: Automatic speech emotion recognition (SER) is a challenging task that plays a crucial role in natural human-computer interaction. One of the main challenges in SER is data scarcity, i.e., insufficient amounts of carefully labeled data to build and fully explore complex deep learning models for emotion classification. This paper aims to address this challenge using a transfer learning strategy combined with spectrogram augmentation. Specifically, we propose a transfer learning approach that leverages a pre-trained residual network (ResNet) model including a statistics pooling layer from speaker recognition trained using large amounts of speaker-labeled data. The statistics pooling layer enables the model to efficiently process variable-length input, thereby eliminating the need for sequence truncation which is commonly used in SER systems. In addition, we adopt a spectrogram augmentation technique to generate additional training data samples by applying random time-frequency masks to log-mel spectrograms to mitigate overfitting and improve the generalization of emotion recognition models. We evaluate the effectiveness of our proposed approach on the interactive emotional dyadic motion capture (IEMOCAP) dataset. Experimental results indicate that the transfer learning and spectrogram augmentation approaches improve the SER performance, and when combined achieve state-of-the-art results.

Xi-Vector Embedding for Speaker Recognition

Abstract: We present a Bayesian formulation for deep speaker embedding, wherein the xi-vector is the Bayesian counterpart of the x-vector, taking into account the uncertainty estimate. On the technology front, we offer a simple and straightforward extension to the now widely used x-vector. It consists of an auxiliary neural net predicting the frame-wise uncertainty of the input sequence. We show that the proposed extension leads to substantial improvement across all operating points, with a significant reduction in error rates and detection cost. On the theoretical front, our proposal integrates the Bayesian formulation of linear Gaussian model to speaker-embedding neural networks via the pooling layer. In one sense, our proposal integrates the Bayesian formulation of the i-vector to that of the x-vector. Hence, we refer to the embedding as the xi-vector, which is pronounced as /zai/ vector. Experimental results on the SITW evaluation set show a consistent improvement of over 17.5% in equal-error-rate and 10.9% in minimum detection cost.

Fine-tuning wav2vec2 for speaker recognition.

Abstract: This paper explores applying the wav2vec2 framework to speaker recognition instead of speech recognition. We study the effectiveness of the pre-trained weights on the speaker recognition task, and how to pool the wav2vec2 output sequence into a fixed-length speaker embedding. To adapt the framework to speaker recognition, we propose a single-utterance classification variant with CE or AAM softmax loss, and an utterance-pair classification variant with BCE loss. Our best performing variant, w2v2-aam, achieves a 1.88% EER on the extended voxceleb1 test set compared to 1.69% EER with an ECAPA-TDNN baseline. Code is available at this https URL.

Convolutional neural network vectors for speaker recognition

Abstract: Deep learning models are now considered state-of-the-art in many areas of pattern recognition. In speaker recognition, several architectures have been studied, such as deep neural networks (DNNs), deep belief networks (DBNs), restricted Boltzmann machines (RBMs), and so on, while convolutional neural networks (CNNs) are the most widely used models in computer vision. The problem is that CNN is limited to the computer vision field due to its structure which is designed for two-dimensional data. To overcome this limitation, we aim at developing a customized CNN for speaker recognition. The goal of this paper is to propose a new approach to extract speaker characteristics by constructing CNN filters linked to the speaker. Besides, we propose new vectors to identify speakers, which we call in this work convVectors. Experiments have been performed with a gender-dependent corpus (THUYG-20 SRE) under three noise conditions : clean, 9db, and 0db. We compared the proposed method with our baseline system and the state-of-the-art methods. Results showed that the convVectors method was the most robust, improving the baseline system by an average of 43%, and recording an equal error rate of 1.05% EER. This is an important finding to understand how deep learning models can be adapted to the problem of speaker recognition.

"Hello, It's Me": Deep Learning-based Speech Synthesis Attacks in the Real World

Abstract: Advances in deep learning have introduced a new wave of voice synthesis tools, capable of producing audio that sounds as if spoken by a target speaker. If successful, such tools in the wrong hands will enable a range of powerful attacks against both humans and software systems (aka machines). This paper documents efforts and findings from a comprehensive experimental study on the impact of deep-learning based speech synthesis attacks on both human listeners and machines such as speaker recognition and voice-signin systems. We find that both humans and machines can be reliably fooled by synthetic speech and that existing defenses against synthesized speech fall short. These findings highlight the need to raise awareness and develop new protections against synthetic speech for both humans and machines.

Large-scale Self-Supervised Speech Representation Learning for Automatic Speaker Verification.

Abstract: The speech representations learned from large-scale unlabeled data have shown better generalizability than those from supervised learning and thus attract a lot of interest to be applied for various downstream tasks. In this paper, we explore the limits of speech representations learned by different self-supervised objectives and datasets for automatic speaker verification (ASV), especially with a well-recognized SOTA ASV model, ECAPA-TDNN [1], as a downstream model. The representations from all hidden layers of the pre-trained model are firstly averaged with learnable weights and then fed into the ECAPA-TDNN as input features. The experimental results on Voxceleb dataset show that the weighted average representation is significantly superior to FBank, a conventional handcrafted feature for ASV. Our best single system achieves 0.564%, 0.561%, and 1.230% equal error rate (EER) on the three official trials of VoxCeleb1, separately. Accordingly, the ensemble system with three pre-trained models can further improve the EER to 0.431%, 0.507% and 1.081%. Among the three evaluation trials, our best system outperforms the winner system [2] of the VoxCeleb Speaker Recognition Challenge 2021 (VoxSRC2021) on the VoxCeleb1-E trial.