Approach For Speaker Recognition Research Articles

Speaker recognition using isomorphic graph attention network based pooling on self-supervised representation

The emergence of self-supervised representation (i.e., wav2vec 2.0) allows speaker-recognition approaches to process spoken signals through foundation models built on speech data. Nevertheless, effective fusion on the representation requires further investigating, due to the inclusion of fixed or sub-optimal temporal pooling strategies. Despite of improved strategies considering graph learning and graph attention factors, non-injective aggregation still exists in the approaches, which may influence the performance for speaker recognition. In this regard, we propose a speaker recognition approach using Isomorphic Graph ATtention network (IsoGAT) on self-supervised representation. The proposed approach contains three modules of representation learning, graph attention, and aggregation, jointly considering learning on the self-supervised representation and the IsoGAT. Then, we perform experiments for speaker recognition tasks on VoxCeleb1&2 datasets, with the corresponding experimental results demonstrating the recognition performance for the proposed approach, compared with existing pooling approaches on the self-supervised representation.

A Cutting-Edge Feature Extraction Approach for Speaker Recognition Leveraging Optimized Variance Spectral Flux and Daubechies Wavelet

A Cutting-Edge Feature Extraction Approach for Speaker Recognition Leveraging Optimized Variance Spectral Flux and Daubechies Wavelet

A Systematic Approach of Advanced Dilated Convolution Network for Speaker Identification

Over the years, the Speaker recognition area is facing various challenges in identifying the speakers accurately. Remarkable changes came into existence with the advent of deep learning algorithms. Deep learning made a remarkable impact on the speaker recognition approaches. This paper introduces a simple novel architectural approach to an advanced Dilated Convolution network. The novel idea is to induce the well-structured log-Melspectrum to the proposed dilated convolution neural network and reduce the number of layers to 11. The network utilizes the Global average pooling to accumulate the outputs from all layers to get the feature vector representation for classification. Only 13 coefficients are extracted per frame of each speech sample. This novel dilated convolution neural network exhibits an accuracy of 90.97%, Equal Error Rate(EER) of 3.75% and 207 Seconds training time outperforms the existing systems on the LibriSpeech corpus.

The effects of face mask on speech production and its implication for forensic speaker identification-A cross-linguistic study.

This study aims to understand the effects of face mask on speech production between Mandarin Chinese and English, and on the automatic classification of mask/no mask speech and individual speakers. A cross-linguistic study on mask speech between Mandarin Chinese and English was then conducted. Continuous speech of the phonetically balanced texts in both Chinese and English versions were recorded from thirty native speakers of Mandarin Chinese (i.e., 15 males and 15 females) with and without wearing a surgical mask. The results of acoustic analyses showed that mask speech exhibited higher F0, intensity, HNR, and lower jitter and shimmer than no mask speech for Mandarin Chinese, whereas higher HNR and lower jitter and shimmer were observed for English mask speech. The results of classification analyses showed that, based on the four supervised learning algorithms (i.e., Linear Discriminant Analysis, Naïve Bayes Classifier, Random Forest, and Support Vector Machine), undesirable performances (i.e., lower than 50%) in classifying the speech with and without a face mask, and highly-variable accuracies (i.e., ranging from 40% to 89.2%) in identifying individual speakers were achieved. These findings imply that the speakers tend to conduct acoustic adjustments to improve their speech intelligibility when wearing surgical mask. However, a cross-linguistic difference in speech strategies to compensate for intelligibility was observed that Mandarin speech was produced with higher F0, intensity, and HNR, while English was produced with higher HNR. Besides, the highly-variable accuracies of speaker identification might suggest that surgical mask would impact the general performance of the accuracy of automatic speaker recognition. In general, therefore, it seems wearing a surgical mask would impact both acoustic-phonetic and automatic speaker recognition approaches to some extent, thus suggesting particular cautions in the real-case practice of forensic speaker identification.

A deep learning approach for text-independent speaker recognition with short utterances

A deep learning approach for text-independent speaker recognition with short utterances

Power Normalized Gammachirp Cepstral (PNGC) coefficients-based approach for robust speaker recognition

Speaker identification or recognition task aims to identify persons from their voices. This paper introduces a new feature extraction approach for robust speaker recognition named Power Normalized Gammachirp Cepstral (PNGC). Key aspect of our method is the use of a biologically motivated auditory perceptual model. For speaker modeling, we use the Gaussian Mixture Model-Universal Background Model (GMM-UBM). The proposed approach includes the following main processing steps: (1) an efficient auditory filter model based on the normalized Gammachirp auditory Filterbank to estimate the cochlea spectral behavior, (2) an environmental noise compensation bloc that employs: a medium-time power analysis, an asymmetric noise-suppression module with a temporal masking module, and a frequency smoothing module to compensate the environmental noise effects, and (3) a power-law nonlinearity. Conducted experimentations on TIMIT and Aurora datasets proved that our proposed PNGC approach achieves an improved recognition accuracy compared to the MFCC and recent PNCC methods for noisy speaker recognition.

Variational autoencoder for prosody‐based speaker recognition

AbstractThis paper describes a novel end‐to‐end deep generative model‐based speaker recognition system using prosodic features. The usefulness of variational autoencoders (VAE) in learning the speaker‐specific prosody representations for the speaker recognition task is examined herein for the first time. The speech signal is first automatically segmented into syllable‐like units using vowel onset points (VOP) and energy valleys. Prosodic features, such as the dynamics of duration, energy, and fundamental frequency ( ), are then extracted at the syllable level and used to train/adapt a speaker‐dependent VAE from a universal VAE. The initial comparative studies on VAEs and traditional autoencoders (AE) suggest that the former can efficiently learn speaker representations. Investigations on the impact of gender information in speaker recognition also point out that gender‐dependent impostor banks lead to higher accuracies. Finally, the evaluation on the NIST SRE 2010 dataset demonstrates the usefulness of the proposed approach for speaker recognition.

An Enhanced Deep Neural Network-Based Approach for Speaker Recognition Using Triumvirate Euphemism Strategy

Automatic Speech Recognition (ASR) has been an intensive research area during the recent years in internet to enable natural human–machine communication. However, the existing Deep Neutral Network (DNN) techniques need more focus on feature extraction process and recognition accuracy. Thus, an enhanced deep neural network (DNN)-based approach for speaker recognition with a novel Triumvirate Euphemism Strategy (TES) is proposed. This overcomes poor feature extraction from Mel-Frequency Cepstral Coefficient (MFCC) map by extracting the features based on petite, hefty and artistry of the features. Then, the features are trained with Silhouette Martyrs Method (SMM) without any inter-class and intra-class separability problems and margins are affixed between classes with three new loss functions, namely A-Loss, AM-Loss and AAM-Loss. Additionally, the parallelization is done by a mini-batch-based BP algorithm in DNN. A novel Frenzied Heap Atrophy (FHA) with a multi-GPU model is introduced in addition with DNN to enhance the parallelized computing that accelerates the training procedures. Thus, the outcome of the proposed technique is highly efficient that provides feasible extraction features and gives incredibly precise results with 97.5% accuracy in the recognition of speakers. Moreover, various parameters were discussed to prove the efficiency of the system and also the proposed method outperformed the existing methods in all aspects.

New Acoustic Features for Synthetic and Replay Spoofing Attack Detection

With the rapid development of intelligent speech technologies, automatic speaker verification (ASV) has become one of the most natural and convenient biometric speaker recognition approaches. However, most state-of-the-art ASV systems are vulnerable to spoofing attack techniques, such as speech synthesis, voice conversion, and replay speech. Due to the symmetry distribution characteristic between the genuine (true) speech and spoof (fake) speech pair, the spoofing attack detection is challenging. Many recent research works have been focusing on the ASV anti-spoofing solutions. This work investigates two types of new acoustic features to improve the performance of spoofing attacks. The first features consist of two cepstral coefficients and one LogSpec feature, which are extracted from the linear prediction (LP) residual signals. The second feature is a harmonic and noise subband ratio feature, which can reflect the interaction movement difference of the vocal tract and glottal airflow of the genuine and spoofing speech. The significance of these new features has been investigated in both the t-stochastic neighborhood embedding space and the binary classification modeling space. Experiments on the ASVspoof 2019 database show that the proposed residual features can achieve from 7% to 51.7% relative equal error rate (EER) reduction on the development and evaluation set over the best single system baseline. Furthermore, more than 31.2% relative EER reduction on both the development and evaluation set shows that the proposed new features contain large information complementary to the source acoustic features.

Using an Automated Speech Recognition Approach to Differentiate Between Normal and Aspirating Swallowing Sounds Recorded from Digital Cervical Auscultation in Children

Use of machine learning to accurately detect aspirating swallowing sounds in children is an evolving field. Previously reported classifiers for the detection of aspirating swallowing sounds in children have reported sensitivities between 79 and 89%. This study aimed to investigate the accuracy of using an automatic speaker recognition approach to differentiate between normal and aspirating swallowing sounds recorded from digital cervical auscultation in children. We analysed 106 normal swallows from 23 healthy children (median 13 months; 52.1% male) and 18 aspirating swallows from 18 children (median 10.5 months; 61.1% male) who underwent concurrent videofluoroscopic swallow studies with digital cervical auscultation. All swallowing sounds were on thin fluids. A support vector machine classifier with a polynomial kernel was trained on feature vectors that comprised the mean and standard deviation of spectral subband centroids extracted from each swallowing sound in the training set. The trained support vector machine was then used to classify swallowing sounds in the test set. We found high accuracy in the differentiation of aspirating and normal swallowing sounds with 98% overall accuracy. Sensitivity for the detection of aspiration and normal swallowing sounds were 89% and 100%, respectively. There were consistent differences in time, power spectral density and spectral subband centroid features between aspirating and normal swallowing sounds in children. This study provides preliminary research evidence that aspirating and normal swallowing sounds in children can be differentiated accurately using machine learning techniques.

Use of Speaker Recognition Approaches for Learning and Evaluating Embedding Representations of Musical Instrument Sounds

Constructing an embedding space for musical instrument sounds that can meaningfully represent new and unseen instruments is important for downstream music generation tasks such as multi-instrument synthesis and timbre transfer. The framework of Automatic Speaker Verification (ASV) provides us with architectures and evaluation methodologies for verifying the identities of unseen speakers, and these can be repurposed for the task of learning and evaluating a musical instrument sound embedding space that can support unseen instruments. Borrowing from state-of-the-art ASV techniques, we construct a musical instrument recognition model that uses a SincNet front-end, a ResNet architecture, and an angular softmax objective function. Experiments on the NSynth and RWC datasets show our model’s effectiveness in terms of equal error rate (EER) for unseen instruments, and ablation studies show the importance of data augmentation and the angular softmax objective. Experiments also show the benefit of using a CQT-based filterbank for initializing SincNet over a Mel filterbank initialization. Further complementary analysis of the learned embedding space is conducted with t-SNE visualizations and probing classification tasks, which show that including instrument family labels as a multi-task learning target can help to regularize the embedding space and incorporate useful structure, and that meaningful information such as playing style, which was not included during training, is contained in the embeddings of unseen instruments.

Evaluating the Performance of Speaker Recognition Solutions in E-Commerce Applications.

Two important tasks in many e-commerce applications are identity verification of the user accessing the system and determining the level of rights that the user has for accessing and manipulating system’s resources. The performance of these tasks is directly dependent on the certainty of establishing the identity of the user. The main research focus of this paper is user identity verification approach based on voice recognition techniques. The paper presents research results connected to the usage of open-source speaker recognition technologies in e-commerce applications with an emphasis on evaluating the performance of the algorithms they use. Four open-source speaker recognition solutions (SPEAR, MARF, ALIZE, and HTK) have been evaluated in cases of mismatched conditions during training and recognition phases. In practice, mismatched conditions are influenced by various lengths of spoken sentences, different types of recording devices, and the usage of different languages in training and recognition phases. All tests conducted in this research were performed in laboratory conditions using the specially designed framework for multimodal biometrics. The obtained results show consistency with the findings of recent research which proves that i-vectors and solutions based on probabilistic linear discriminant analysis (PLDA) continue to be the dominant speaker recognition approaches for text-independent tasks.

A deep learning approach for speaker recognition

Speaker verification (SV) is an important branch in speaker recognition. Several approaches have been investigated within the last few decades. In this context, deep learning has received much more interest by speech processing researchers, and it was introduced recently in speaker recognition. In most cases, deep learning models are adapted from speech recognition applications and applied to speaker recognition, and they have been showing their capability of being competitors to the state-of-the-art approaches. Nevertheless, the use of deep learning in speaker recognition is still linked to speech recognition. In this study, we are proposing a new way to use deep neural networks (DNNs) in speaker recognition, in the purpose to facilitate to DNN to learn features distribution. We have been motivated by our previous work, where we have proposed a novel scoring method that works perfectly with clean speech, but it needs improvements under noisy conditions. For this reason, we are aiming to transform the extracted feature vectors (MFCCs) into enhanced feature vectors, that we denote Deep Speaker Features (DeepSFs). Experiments have been conducted on THUYG-20 SRE corpus, and significant results have been achieved. Moreover, this new method outperformed both i-vector/PLDA and our baseline system in both clean and noisy conditions.

A framework for speaker retrieval and identification through unsupervised learning

Speaker recognition is a task of remarkable relevance, with applications in diversified domains. Recently, mainly due to the facilities in audio-visual content acquisition, the capacity of analyzing growing datasets independent of labeled data has become a crucial advantage. This paper presents a speaker recognition approach based on recent unsupervised learning methods, which do not require any labeled data or user intervention. The approach is organized in terms of a framework which exploits a rank-based formulation. The similarity information defined by speaker modeling techniques is encoded in ranked lists, which are used as input by the unsupervised learning algorithms. Vector quantization, Gaussian mixture models and i-vectors are employed as modeling techniques, while the algorithms RL-Sim and ReckNN are used for unsupervised learning tasks. The framework was experimentally evaluated on query-by-example speaker retrieval and speaker identification tasks, both on clean and noisy speech recordings. An experimental evaluation was conducted on three public datasets, different languages, and recordings conditions. Effectiveness gains up to +56% on retrieval measures were obtained through the use of unsupervised learning algorithms over traditional speaker recognition techniques.

HMM-Based Phrase-Independent i-Vector Extractor for Text-Dependent Speaker Verification

The low-dimensional i-vector representation of speech segments is used in the state-of-the-art text-independent speaker verification systems. However, i-vectors were deemed unsuitable for the text-dependent task, where simpler and older speaker recognition approaches were found more effective. In this work, we propose a straightforward hidden Markov model (HMM) based extension of the i-vector approach, which allows i-vectors to be successfully applied to text-dependent speaker verification. In our approach, the Universal Background Model (UBM) for training phrase-independent i-vector extractor is based on a set of monophone HMMs instead of the standard Gaussian Mixture Model (GMM). To compensate for the channel variability, we propose to precondition i-vectors using a regularized variant of within-class covariance normalization, which can be robustly estimated in a phrase-dependent fashion on the small datasets available for the text-dependent task. The verification scores are cosine similarities between the i-vectors normalized using phrase-dependent s-norm. The experimental results on RSR2015 and RedDots databases confirm the effectiveness of the proposed approach, especially in rejecting test utterances with a wrong phrase. A simple MFCC based i-vector/HMM system performs competitively when compared to very computationally expensive DNN-based approaches or the conventional relevance MAP GMM-UBM, which does not allow for compact speaker representations. To our knowledge, this paper presents the best published results obtained with a single system on both RSR2015 and RedDots dataset.

A PLDA approach for language and text independent speaker recognition

There are many factors affecting the variability of an i-vector extracted from a speech segment such as the acoustic content, segment duration, handset type and background noise. The language being spoken is one of the sources of variation which has received limited focus due to the lack of multilingual resources available. Consequently, the discrimination performance is much lower under multilingual trial condition. Standard session-compensation techniques such as Within-Class Covariance Normalization (WCCN), Linear Discriminant Analysis (LDA) and Probabilistic LDA (PLDA) cannot robustly compensate for language source of variation as the amount of data is limited to represent such variability. Source normalization technique which was developed to compensate for speech-source-variation, offered superior performance in cross-language trials by providing better estimation of within-speaker scatter matrix in WCCN and LDA techniques. However, neither language normalization nor the state-of-the-art PLDA algorithm is capable of modeling language variability on a dataset with insufficient multilingual utterances for each speaker, resulting in a poor performance in cross-language trial condition.This study is an extension to our initial developments of a language-independent PLDA training algorithm which aimed at reducing the effect of language as a source of variability on the performance of speaker recognition. We will provide a thorough analysis of how the proposed approach can utilize multilingual training data from bilingual speakers to robustly compensate for the effect of languages. Evaluated on multilingual trial condition, the proposed solution demonstrated over 10% EER and 13% minimum DCF relative improvement on NIST 2008 speaker recognition evaluation as well as 12.4% EER and 23% minimum DCF on PRISM evaluation set over the baseline system while also providing improvement in other trial conditions.

Program Guardian: screening system with a novel speaker recognition approach for smart TV

This paper presents Program Guardian, which is a speaker recognition-based screening system for smart TV. The system identifies a specific person from his or her voice such that the smart TV can provide suitable programs for that person. This system is based on a robust speaker recognition system that uses robust principal component analysis (RPCA) and a sparse representation classifier (SRC). First, i-vectors that are generated from supervectors of Gaussian mixture models (GMMs) are used to generate the basic atoms of an over-complete dictionary. The i-vectors are then transformed using RPCA. The SRC is produced from transformed i-vector-based RPCA vectors. Finally, the sparse representation classifier corresponding to the target speaker with the least reconstruction error is constructed. NIST speaker recognition evaluation data base is used in our experiment. The results show that the proposed speaker recognition system is feasible and offers advantages over accuracy.

Use of Silence as an Altered Approach for Speaker Recognition

Speaker recognition is an important application of Speech Signal Processing and has been used in public safety, authenticating users for important financial transactions, access control systems and many more. The conventional approach in speaker identification and speaker verification has been to remove the silence from the recorded speech signal and further extract the significant features from the residual signal for recognition. This paper presents an alternative approach and puts forth the experimental results of obtaining silence as a parameter to check if the pattern of pauses/silence for train and test files recorded for individual speaker match. The paper emphasizes the approach in which a paragraph is recorded for 8 speakers and used as train files. The duration of silence/pauses of the speaker in a paragraph are obtained. This silence obtained is compared with the silence obtained from test file the matching of pattern of the silences decides the identity of the speaker.

An Improved Ranking-Based Feature Enhancement Approach for Robust Speaker Recognition

Although the field of automatic speaker or speech recognition has been extensively studied over the past decades, the lack of robustness has remained a major challenge. Feature warping is a promising approach and its effectiveness significantly depends on the relative positions of each of the features in a sliding window. However, the relative positions are changed due to the non-linear effect of noise. Aiming at the problem, this paper takes the advantage of ranking feature, which is obtained directly by sorting a feature sequence in descending order, to propose a method. It first labels the central frame in a sliding window as speech or noise dominant (“reliable” or “unreliable”). In the unreliable case, the ranking of the central frame is estimated. Subsequently, the estimated ranking is mapped to a warped feature using a desired target distribution for recognition experiments. Through the theoretical analysis and experimental results, it is found that autocorrelation of a ranking sequence is larger than that of the corresponding feature sequence. What is more, rank correlation is not easily influenced by abnormal data or data that are highly variable. Thus, this paper deals with a ranking sequence rather than a feature sequence. The proposed feature enhancement approach is evaluated in an open-set speaker recognition system. The experimental results show that it outperforms missing data method based on linear interpolation and feature warping in terms of recognition performance in all noise conditions. Furthermore, the method proposed here is a feature-based method, which may be combined with other technologies, such as model-based, scores-based, to enhance the robustness of speaker or speech recognition system.

Fully automated speaker identification and intelligibility assessment in dysarthria disease using auditory knowledge

Millions of children and adults suffer from acquired or congenital neuro-motor communication disorders that can affect their speech intelligibility. The automatically characterization of speech impairment can contribute to improve the patient's life quality, and assist experts in assessment and treatment design. In this paper, we present new approaches to improve the analysis and classification of disordered speech. First, we propose an automatic speaker recognition approach especially adapted to identify dysarthric speakers. Secondly, we suggest a method for the automatic assessment of the dysarthria severity level. For this purpose, a model simulating the external, middle and inner parts of the ear is presented. This ear model provides relevant auditory-based cues that are combined with the usual Mel-Frequency Cepstral Coefficients (MFCC) to represent atypical speech utterances. The experiments are carried out by using data of both Nemours and Torgo databases of dysarthric speech. Gaussian Mixture Models (GMMs), Support Vector Machines (SVMs) and hybrid GMM/SVM systems are tested and compared in the context of dysarthric speaker identification and assessment. The experimental results achieve a correct speaker identification rate of 97.2% which can be considered promising for this novel approach; also the existing assessment systems are outperformed with a 93.2% correct classification rate of dysarthria severity levels.