Speaker Identification Accuracy Research Articles

Enhancing Speaker Recognition Models with Noise-Resilient Feature Optimization Strategies

This paper delves into an in-depth exploration of speaker recognition methodologies, with a primary focus on three pivotal approaches: feature-level fusion, dimension reduction employing principal component analysis (PCA) and independent component analysis (ICA), and feature optimization through a genetic algorithm (GA) and the marine predator algorithm (MPA). This study conducts comprehensive experiments across diverse speech datasets characterized by varying noise levels and speaker counts. Impressively, the research yields exceptional results across different datasets and classifiers. For instance, on the TIMIT babble noise dataset (120 speakers), feature fusion achieves a remarkable speaker identification accuracy of 92.7%, while various feature optimization techniques combined with K nearest neighbor (KNN) and linear discriminant (LD) classifiers result in a speaker verification equal error rate (SV EER) of 0.7%. Notably, this study achieves a speaker identification accuracy of 93.5% and SV EER of 0.13% on the TIMIT babble noise dataset (630 speakers) using a KNN classifier with feature optimization. On the TIMIT white noise dataset (120 and 630 speakers), speaker identification accuracies of 93.3% and 83.5%, along with SV EER values of 0.58% and 0.13%, respectively, were attained utilizing PCA dimension reduction and feature optimization techniques (PCA-MPA) with KNN classifiers. Furthermore, on the voxceleb1 dataset, PCA-MPA feature optimization with KNN classifiers achieves a speaker identification accuracy of 95.2% and an SV EER of 1.8%. These findings underscore the significant enhancement in computational speed and speaker recognition performance facilitated by feature optimization strategies.

Optimizing Speaker Identification through SincsquareNet and SincNet Fusion with Attention Mechanism

Advancements in machine learning and deep learning benefit access control systems, forensics, and biometrics particularly in speaker identification systems. The SincNet architecture is a distinct convolutional neural network (CNN) designed for speaker identification. It operates by taking one-dimensional raw speech and directing it through its initial convolutional layer, consisting of Sinc filters. In this work, we present a SincsquareNet, a CNN that efficiently learns customized triangular band-pass filters using trainable Sinc-squared functions. Also, we propose the fusion of SincsqaureNet with SincNet for robust speaker identification. Further, a self-attention mechanism is employed to obtain discriminating features. The Librispeech dataset is used to validate the proposed framework. This approach makes use of the best of both filters, helps the network learn more robust features, and converges faster. The findings of the experiments show that, when compared to SincsquareNet, speaker identification accuracy was relatively improved by 8%, while validation loss was reduced by 7%.

Siamese Networks for Speaker Identification on Resource-Constrained Platforms

This paper investigates the implementation of a lightweight Siamese neural network for enhancing speaker identification accuracy and inference speed in embedded systems. Integrating speaker identification into embedded systems can improve portability and versatility. Siamese neural networks achieve speaker identification by comparing input voice samples to reference voices in a database, effectively extracting features and classifying speakers accurately. Considering the trade-off between accuracy and complexity, as well as hardware constraints in embedded systems, various neural networks could be applied to speaker identification. This paper compares the incorporation of CNN architectures targeted for embedded systems, MCUNet, SqueezeNet and MobileNetv2, to implement Siamese neural networks on a Raspberry Pi. Our experiments demonstrate that MCUNet achieves 85% accuracy with a 0.23-second inference time. In comparison, the larger MobileNetv2 attains 84.5% accuracy with a 0.32-second inference time. Additionally, contrastive loss was superior to binary cross-entropy loss in the Siamese neural network. The system using contrastive loss had almost 68% lower loss scores, resulting in a more stable performance and more accurate predictions. In conclusion, this paper establishes that an appropriate lightweight Siamese neural network, combined with contrastive loss, can significantly improve speaker identification accuracy, and enable efficient deployment on resource-constrained platforms.

Non-uniform Speaker Disentanglement For Depression Detection From Raw Speech Signals.

While speech-based depression detection methods that use speaker-identity features, such as speaker embeddings, are popular, they often compromise patient privacy. To address this issue, we propose a speaker disentanglement method that utilizes a non-uniform mechanism of adversarial SID loss maximization. This is achieved by varying the adversarial weight between different layers of a model during training. We find that a greater adversarial weight for the initial layers leads to performance improvement. Our approach using the ECAPA-TDNN model achieves an F1-score of 0.7349 (a 3.7% improvement over audio-only SOTA) on the DAIC-WoZ dataset, while simultaneously reducing the speaker-identification accuracy by 50%. Our findings suggest that identifying depression through speech signals can be accomplished without placing undue reliance on a speaker's identity, paving the way for privacy-preserving approaches of depression detection.

Text-Independent Speaker Recognition System Using Feature-Level Fusion for Audio Databases of Various Sizes

To improve the speaker recognition rate, we propose a speaker recognition model based on the fusion of different kinds of speech features. A new type of feature aggregation methodology with a total of 18 features is proposed and includes mel frequency cepstral coefficient (MFCC), linear predictive coding (LPC), perceptual linear prediction (PLP), root mean square (RMS), centroid, and entropy features along with their delta (Δ) and delta–delta (ΔΔ) feature vectors. The proposed approach is tested on five different sizes of speech datasets, namely the NIST-2008, voxforge, ELSDSR, VCTK, and voxceleb1 speech corpora. The results are evaluated using the MATLAB classification learner application with the linear discriminant (LD), K nearest neighbor (KNN), and ensemble classifiers. For the NIST-2008 and voxforge datasets, the best SI accuracy of 96.9% and 100% and the lowest speaker verification (SV) equal error rate (EER) values of 0.2% and 0% are achieved with the LD and KNN classifiers, respectively. For the VCTK and ELSDSR datasets, the best SI accuracy of 100% and the lowest SV EER of 0% are achieved with all three classifiers using different feature-level fusion approaches, while the highest SI accuracy and lowest EER achieved on the voxceleb1 database are 90% and 4.07%, respectively, using the KNN classifier. From the experimental results, it is observed that the fusion of different features with their delta and delta–delta values shows an increase in speaker identification accuracy of 10–50%, and the EER value for SV is reduced compared to the value obtained with a single feature.

Speaker Verification Based on Single Channel Speech Separation

In multi-speaker scenarios, speech processing tasks like speaker identification and speech recognition are susceptible to noise and overlapped voices. As the overlapped voices are a complicated mixture of signals, a target extraction method from this mixture is a good front end solution for further processing like understanding and classifying. The quality of speech separation can be assessed by the noise ratio or subjective scoring and can also be assessed by accuracy of the downstream processing tasks like speaker identification. In order to make the separation model and speaker identification model more adapted to complex multi-speaker speech overlapping scenarios, this research investigates the speech separation model and incorporate with a voiceprint recognition task. This paper proposes a feature-scale single channel speech separation network connected to a back end speaker verification network with MFCCT feature, so the accuracy of speaker identification indicates the quality of speech separation task. The datasets are prepared by synthesizing Voxceleb1 data, and used for training and testing. The results show that using an objective downstream evaluation can effectively improve the overall performance, as the optimized speech separation model significantly reduced the error rate of speaker verification.

FEATURES OF THE IMPLEMENTATION OF THE SPEAKER IDENTIFICATION SOFTWARE SYSTEM

The proposed architecture of the identification software system in the form of class and sequence diagrams. The main criteria for assessing the accuracy of speaker identification were studied and possible sources of loss of speaker identification accuracy were identified, which can be used when building a speaker identification system. A software system based on the proposed architecture and previously developed identification algorithms and methods was created.
 The following conclusions can be drawn on the basis of the performed research: approaches to the construction of existing announcer identification systems are considered; the main criteria for assessing the accuracy of announcer identification were investigated and the main sources of loss of accuracy during announcer identification were identified; the structural construction of the announcer identification system is considered, taking into account the identified sources of loss of accuracy during announcer identification; the proposed architecture of the speaker identification system in the UML language in the form of class and sequence diagrams; a software system was built that implements the functions of speech signal identification according to the methods and algorithm proposed in previous works.
 The software system uses a ranking method based on three different criteria. These include: calculation of the proximity of two-dimensional probability density function curves for the frequency of the main tone and the location in the spectrum of three frequency ranges that are extracted from the speech recorded in the speech signal; calculation of the proximity of the probability density function curves for each of these features separately; calculation of the degree of closeness of the absolute maxima of the formant spectra extracted from the speech recorded in the speech signal.

Contrastive Adversarial Domain Adaptation Networks for Speaker Recognition.

Domain adaptation aims to reduce the mismatch between the source and target domains. A domain adversarial network (DAN) has been recently proposed to incorporate adversarial learning into deep neural networks to create a domain-invariant space. However, DAN's major drawback is that it is difficult to find the domain-invariant space by using a single feature extractor. In this article, we propose to split the feature extractor into two contrastive branches, with one branch delegating for the class-dependence in the latent space and another branch focusing on domain-invariance. The feature extractor achieves these contrastive goals by sharing the first and last hidden layers but possessing decoupled branches in the middle hidden layers. For encouraging the feature extractor to produce class-discriminative embedded features, the label predictor is adversarially trained to produce equal posterior probabilities across all of the outputs instead of producing one-hot outputs. We refer to the resulting domain adaptation network as "contrastive adversarial domain adaptation network (CADAN)." We evaluated the embedded features' domain-invariance via a series of speaker identification experiments under both clean and noisy conditions. Results demonstrate that the embedded features produced by CADAN lead to a 33% improvement in speaker identification accuracy compared with the conventional DAN.

Exploring racial and gender disparities in voice biometrics

Systemic inequity in biometrics systems based on racial and gender disparities has received a lot of attention recently. These disparities have been explored in existing biometrics systems such as facial biometrics (identifying individuals based on facial attributes). However, such ethical issues remain largely unexplored in voice biometric systems that are very popular and extensively used globally. Using a corpus of non-speech voice records featuring a diverse group of 300 speakers by race (75 each from White, Black, Asian, and Latinx subgroups) and gender (150 each from female and male subgroups), we explore and reveal that racial subgroup has a similar voice characteristic and gender subgroup has a significant different voice characteristic. Moreover, non-negligible racial and gender disparities exist in speaker identification accuracy by analyzing the performance of one commercial product and five research products. The average accuracy for Latinxs can be 12% lower than Whites (p < 0.05, 95% CI 1.58%, 14.15%) and can be significantly higher for female speakers than males (3.67% higher, p < 0.05, 95% CI 1.23%, 11.57%). We further discover that racial disparities primarily result from the neural network-based feature extraction within the voice biometric product and gender disparities primarily due to both voice inherent characteristic difference and neural network-based feature extraction. Finally, we point out strategies (e.g., feature extraction optimization) to incorporate fairness and inclusive consideration in biometrics technology.

Attention based gender and nationality information exploration for speaker identification

Gender and nationality information has not been exploited in large-scale speaker recognition despite being provided in the popular VoxCeleb1 dataset. This paper explores methods that combine high-level features extracted from the gender and nationality information with low-level acoustic features for speaker identification. To our knowledge, this is the first time that the gender and nationality information provided in VoxCeleb1 is utilized in speaker identification. Specifically, we propose Gender-Guided Spectrogram-Attention network and Nationality-Guided Spectrogram-Attention network that embed gender and nationality information into the spectrogram features, respectively. The resulting gender and nationality embeddings are then used with the spectrogram features together for classification. Experimental results show that the proposed methods can successfully capture the gender and nationality information of the speakers, and can effectively improve speaker identification accuracy.

CASA-based speaker identification using cascaded GMM-CNN classifier in noisy and emotional talking conditions

This work aims at intensifying text-independent speaker identification performance in real application situations such as noisy and emotional talking conditions. This is achieved by incorporating two different modules: a Computational Auditory Scene Analysis (CASA) based pre-processing module for noise reduction and “cascaded Gaussian Mixture Model – Convolutional Neural Network (GMM-CNN) classifier for speaker identification” followed by emotion recognition. This research proposes and evaluates a novel algorithm to improve the accuracy of speaker identification in emotional and highly-noise susceptible conditions. Experiments demonstrate that the proposed model yields promising results in comparison with other classifiers when “Speech Under Simulated and Actual Stress (SUSAS) database, Emirati Speech Database (ESD), the Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS)” database and the “Fluent Speech Commands” database are used in a noisy environment.

Speaker identification based on Radon transform and CNNs in the presence of different types of interference for Robotic Applications

Both automatic speaker identification (ASI) and speech recognition can be utlized now for the control of modern robots. An ASI algorithm can be implemented at a speech interface of the robot to determine the identity of the person allowed to deal with the robot, while speech recognition can be implemented for the interpretation of the order given to the robot. Robustness of the ASI system is a challenging task in the presence of speech degradations such as noise and interference. This study presents a new approach to improve the accuracy of speaker identification in the presence of interference for robot control applications with a convolutional neural network (CNN). First, the speech signal from the speaker is divided into segments, each of which is transformed into a spectrogram, and hence Radon transformation is estimated for this spectrogram. The spectrogram resolves the speech segment into a map of power distribution with both time and frequency. Together, the spectrograms and their Radon transforms are used as inputs to a proposed CNN-based deep learning model. Necessary refinements are undertaken and the resulting optimized “Radon-Deep-Learning Model (RDLM) is compared with a benchmark model. The proposed model consists of six convolutional (CNV) layers followed by six Max. pooling layers, while the benchmark model consists of three CNV layers followed by three Max. pooling layers. Experimental results reveal that the proposed RDLM model achieves a high classification accuracy up to 97.5%, which is more than double the performance reported for some traditional methods that are used for speaker identification.

Comparisons of extreme learning machine and backpropagation-based i-vector approach for speaker identification

The extreme learning machine (ELM) is one of the machine learning applications used for regression and classification systems. In this paper, an extended comparison between an ELM and the backpropagation neural network (BPNN)-based i-vector is given in terms of a closed-set speaker identification task using 120 speakers from the TIMIT database. The system is composed of the mel frequency cepstal coefficient (MFCC) and power normalized cepstal coefficient (PNCC) approaches to form the feature extraction stage, while the cepstral mean variance normalization (CMVN) and feature warping are applied in order to mitigate the linear channel effect. The system is utilized with equal numbers of speakers of both genders with 120 speakers with eight dialects from the TIMIT database. The results demonstrate that the combination of the i-vector with the ELM for different features has the highest speaker identification accuracy (SIA) compared with the combination of the BPNN with the i-vector. The results also show that the i-vector with ELM approach is faster than the BPNN-based i-vector and it has the highest SIA.

Comparison of feature extraction and normalization methods for speaker recognition using grid-audiovisual database

&lt;p&gt;&lt;span lang="EN-GB"&gt;In this paper, different feature extraction and feature normalization methods are investigated for speaker recognition. With a view to give a good representation of acoustic speech signals, Power Normalized Cepstral Coefficients (PNCCs) and Mel Frequency Cepstral Coefficients (MFCCs) are employed for feature extraction. Then, to mitigate the effect of linear channel, Cepstral Mean-Variance Normalization (CMVN) and feature warping are utilized. The current paper investigates Text-independent speaker identification system by using 16 coefficients from both the MFCCs and PNCCs features. Eight different speakers are selected from the GRID-Audiovisual database with two females and six males. The speakers are modeled using the coupling between the Universal Background Model and Gaussian Mixture Models (GMM-UBM) in order to get a fast scoring technique and better performance. The system shows 100% in terms of speaker identification accuracy. The results illustrated that PNCCs features have better performance compared to the MFCCs features to identify females compared to male speakers. Furthermore, feature wrapping reported better performance compared to the CMVN method. &lt;/span&gt;&lt;/p&gt;

A network model of speaker identification with new feature extraction methods and asymmetric BLSTM

Speaker identification has recently attracted considerable attention in speaker recognition. Environmental noise and short utterance pose two challenges for accurate speaker identification. In this paper, a network model with new feature extraction methods and a new bi-directional long short-term memory network is proposed to identify the speaker. Specifically, this paper proposes to combine the mel-spectrogram and cochleagram to generate two new features, named MC-spectrogram and MC-cube. They have stronger robustness and can obtain more abundant voiceprint feature in the short utterance. Then, multi-dimensional CNNs are applied to process MC-spectrogram and MC-cube features correspondingly. They contain multi-dimensional convolution kernels, which can learn the voiceprint features more efficiently. In addition, the context information is ignored by CNN. And the forward voiceprint features are more crucial because the voiceprint features concentrate on the back part in the short utterance. Asymmetric bi-directional long short-time memory network (ABLSTM) is proposed to further learn the voiceprint features in global feature learning. It can improve the accuracy of speaker identification. According to the diverse dimension of input, the proposed network model can manifest diverse patterns, which are named Audio-1DCNN-ABLSTM, MCS(MC-spectrogram)-2DCNN-ABLSTM and MCC(MC-cube)-3DCNN-ABLSTM. From the experimental results, it is shown that the diverse patterns can achieve superior accuracy and robustness in the short utterance with extra environmental noise. Furthermore, the proposed network model provides a reliable solution in text-independent speaker identification.

Text-Independent Speaker Identification Through Feature Fusion and Deep Neural Network

Speaker identification refers to the process of recognizing human voice using artificial intelligence techniques. Speaker identification technologies are widely applied in voice authentication, security and surveillance, electronic voice eavesdropping, and identity verification. In the speaker identification process, extracting discriminative and salient features from speaker utterances is an important task to accurately identify speakers. Various features for speaker identification have been recently proposed by researchers. Most studies on speaker identification have utilized short-time features, such as perceptual linear predictive (PLP) coefficients and Mel frequency cepstral coefficients (MFCC), due to their capability to capture the repetitive nature and efficiency of signals. Various studies have shown the effectiveness of MFCC features in correctly identifying speakers. However, the performances of these features degrade on complex speech datasets, and therefore, these features fail to accurately identify speaker characteristics. To address this problem, this study proposes a novel fusion of MFCC and time-based features (MFCCT), which combines the effectiveness of MFCC and time-domain features to improve the accuracy of text-independent speaker identification (SI) systems. The extracted MFCCT features were fed as input to a deep neural network (DNN) to construct the speaker identification model. Results showed that the proposed MFCCT features coupled with DNN outperformed existing baseline MFCC and time-domain features on the LibriSpeech dataset. In addition, DNN obtained better classification results compared with five machine learning algorithms that were recently utilized in speaker recognition. Moreover, this study evaluated the effectiveness of one-level and two-level classification methods for speaker identification. The experimental results showed that two-level classification presented better results than one-level classification. The proposed features and classification model for identifying a speaker can be widely applied to different types of speaker datasets.

Speaker Anonymization for Personal Information Protection Using Voice Conversion Techniques

As speech-based user interfaces integrated in the devices such as AI speakers become ubiquitous, a large amount of user voice data is being collected to enhance the accuracy of speech recognition systems. Since such voice data contain personal information that can endanger the privacy of users, the issue of privacy protection in the speech data has garnered increasing attention after the introduction of the General Data Protection Regulation in the EU, which implies that restrictions and safety measures for the use of speech data become essential. This study aims to filter the speaker-related voice biometrics present in speech data such as voice fingerprint without altering the linguistic content to preserve the usefulness of the data while protecting the privacy of users. To achieve this, we propose an algorithm that produces anonymized speeches by adopting many-to-many voice conversion techniques based on variational autoencoders (VAEs) and modifying the speaker identity vectors of the VAE input to anonymize the speech data. We validated the effectiveness of the proposed method by measuring the speaker-related information and the original linguistic information retained in the resultant speech, using an open source speaker recognizer and a deep neural network-based automatic speech recognizer, respectively. Using the proposed method, the speaker identification accuracy of the speech data was reduced to 0.1-9.2%, indicating successful anonymization, while the speech recognition accuracy was maintained as 78.2-81.3%.

Thorough evaluation of TIMIT database speaker identification performance under noise with and without the G.712 type handset

In this work, a speaker identification system is proposed which employs two feature extraction models, namely: the power normalized cepstral coefficients and the mel frequency cepstral coefficients. Both features are subjected to acoustic modeling using a Gaussian mixture model–universal background model. The purpose of this work is to provide a thorough evaluation of the effect of different types of noise on the speaker identification accuracy (SIA) and thereby providing benchmark figures for future comparative studies. In particular, the additive white Gaussian noise and eight non-stationary noise types (with and without the G.712 type handset) corresponding to various signal to noise ratios are tested. Fusion strategies are also employed using late fusion methods: maximum, weighted sum, and mean fusion. The measurements of randomly selected 120 speakers from the TIMIT database are employed and the SIA is used to measure the system performance. The weighted sum fusion resulted in the best performance in terms of SIA with noisy speech. The proposed model given in this work and its related analysis paves the way for further work in this important area.

Speaker identification based on normalized pitch frequency and Mel Frequency Cepstral Coefficients

This paper presents an efficient approach for automatic speaker identification based on cepstral features and the Normalized Pitch Frequency (NPF). Most relevant speaker identification methods adopt a cepstral strategy. Inclusion of the pitch frequency as a new feature in the speaker identification process is expected to enhance the speaker identification accuracy. In the proposed framework for speaker identification, a neural classifier with a single hidden layer is used. Different transform domains are investigated for reliable feature extraction from the speech signal. Moreover, a pre-processing noise reduction step, is used prior to the feature extraction process to enhance the performance of the speaker identification system. Simulation results prove that the NPF as a feature in speaker identification enhances the performance of the speaker identification system, especially with the Discrete Cosine Transform (DCT) and wavelet denoising pre-processing step.

Evaluation of a speaker identification system with and without fusion using three databases in the presence of noise and handset effects

In this study, a speaker identification system is considered consisting of a feature extraction stage which utilizes both power normalized cepstral coefficients (PNCCs) and Mel frequency cepstral coefficients (MFCC). Normalization is applied by employing cepstral mean and variance normalization (CMVN) and feature warping (FW), together with acoustic modeling using a Gaussian mixture model-universal background model (GMM-UBM). The main contributions are comprehensive evaluations of the effect of both additive white Gaussian noise (AWGN) and non-stationary noise (NSN) (with and without a G.712 type handset) upon identification performance. In particular, three NSN types with varying signal to noise ratios (SNRs) were tested corresponding to street traffic, a bus interior, and a crowded talking environment. The performance evaluation also considered the effect of late fusion techniques based on score fusion, namely, mean, maximum, and linear weighted sum fusion. The databases employed were TIMIT, SITW, and NIST 2008; and 120 speakers were selected from each database to yield 3600 speech utterances. As recommendations from the study, mean fusion is found to yield overall best performance in terms of speaker identification accuracy (SIA) with noisy speech, whereas linear weighted sum fusion is overall best for original database recordings.