Automatic Speaker Identification Research Articles

Automatic speaker and age identification of children from raw speech using sincNet over ERB scale

This paper presents the newly developed non-native children’s English speech (NNCES) corpus to reveal the findings of automatic speaker and age recognition from raw speech. Convolutional neural networks (CNN), which have the ability to learn low-level speech representations, can be fed directly with raw speech signals instead of using traditional hand-crafted features. Moreover, the filters that were learned using standard CNNs appeared to be noisy because they consider all elements of each filter. In contrast, sincNet can be able to generate more meaningful filters simply by replacing the first convolutional layer by a sinc-layer in standard CNNs. The low and high cutoff frequencies of the rectangular band-pass filter are the only parameters that can be learned in sincNet, which has the potential to extract significant speech cues from the speaker, such as pitch and formants. In this work, the sincNet model is significantly changed by switching from baseline Mel scale initializations to equivalent rectangular bandwidth (ERB) initializations, which has the added benefit of allocating additional filters in the lower region of the spectrum. Additionally, it needs to be highlighted that the novel sincNet model is well suited to identify the age of the children. The investigations on both read and spontaneous speech tasks in speaker identification, gender independent & dependent age-group identification of children outperform the baseline models with varying relative improvements in terms of accuracy.

Quran reciter identification using NASNetLarge

AbstractSpeaker identification has significant advantages for the field of human–computer interaction. Recently, many scholars have made contributions in this field and successfully created deep learning models for automatic speaker identification systems. However, most of the speech signal processing work is limited to English-only applications, despite numerous challenges with Arabic speech, particularly with the recitation of the Holy Quran, which is the Islamic holy book. In the light of these considerations, this study proposes a model for identifying the reciter of the Holy Quran using a dataset of 11,000 audio samples extracted from 20 Quran reciters. To enable feeding the audio samples' visual representation to the pre-trained models, the audio samples are converted from their original audio representation to visual representation using the Mel-Frequency Cepstrum Coefficients. Six pre-trained deep learning models are evaluated separately in the proposed model. The results from the test dataset reveal that the NASNetLarge model achieved the highest accuracy rate of 98.50% among the pre-trained models used in this study.

A Novel Convolutional Neural Network Model for Automatic Speaker Identification From Speech Signals

A Novel Convolutional Neural Network Model for Automatic Speaker Identification From Speech Signals

An efficient Automatic Speaker Identification System based Pitch Frequency Estimation in Degraded Environmental Conditions

An efficient Automatic Speaker Identification System based Pitch Frequency Estimation in Degraded Environmental Conditions

Improving the Performance of Low-resourced Speaker Identification with Data Preprocessing

Automatic speaker identification is done to tackle daily security problems. Speech data collection is an essential but very challenging task for under-resourced languages like Burmese. The speech quality is crucial to accurately recognize the speaker’s identity. This work attempted to find the optimal speech quality appropriate for Burmese tone to enhance identification compared with other more richy resourced languages on Mel-frequency cepstral coefficients (MFCCs). A Burmese speech dataset was created as part of our work because no appropriate dataset available for use. In order to achieve better performance, we preprocessed the foremost recording quality proper for not only Burmese tone but also for nine other Asian languages to achieve multilingual speaker identification. The performance of the preprocessed data was evaluated by comparing with the original data, using a time delay neural network (TDNN) together with a subsampling technique that can reduce time complexity in model training. The experiments were investigated and analyzed on speech datasets of ten Asian languages to reveal the effectiveness of the data preprocessing. The dataset outperformed the original dataset with improvements in terms of equal error rate (EER). The evaluation pointed out that the performance of the system with the preprocessed dataset improved that of the original dataset.

Generative Adversarial Network with Convolutional Wavelet Packet Transforms for Automated Speaker Recognition and Classification

Speech is an effective mode of communication that always conveys abundant and pertinent information, such as the gender, accent, and other distinguishing characteristics of the speaker. These distinctive characteristics allow researchers to identify human voices using artificial intelligence (AI) techniques, which are useful for forensic voice verification, security and surveillance, electronic voice eavesdropping, mobile banking, and mobile purchasing. Deep learning (DL) and other advances in hardware have piqued the interest of researchers studying automatic speaker identification (SI). In recent years, Generative Adversarial Networks (GANs) have demonstrated exceptional ability in producing synthetic data and improving the performance of several machine learning tasks. The capacity of Convolutional Wavelet Packet Transform (CWPT) and Generative Adversarial Networks are combined in this paper to propose a novel way of enhancing the accuracy and robustness of Speaker Recognition and Classification systems. Audio signals are dissected using the Convolutional Wavelet Packet Transform into a multi-resolution, time-frequency representation that faithfully preserves local and global characteristics. The improved audio features better precisely describe speech traits and handle pitch, tone, and pronunciation variations that are frequent in speaker recognition tasks. Using GANs to create synthetic speech samples, our suggested method GAN-CWPT enriches the training data and broadens the dataset's diversity. The generator and discriminator components of the GAN architecture have been tweaked to produce realistic speech samples with attributes quite similar to genuine speaker utterances. The new dataset enhances the Speaker Recognition and Classification system's robustness and generalization, even in environments with little training data. We conduct extensive tests on standard speaker recognition datasets to determine how well our method works. The findings demonstrate that, compared to conventional methods, the GAN-CWPTs combination significantly improves speaker recognition, classification accuracy, and efficiency. Additionally, the suggested model GAN-CWPT exhibits stronger generalization on unknown speakers and excels even with loud and poor audio inputs.

Evaluating the effects of task design on unfamiliar Francophone listener and automatic speaker identification performance

Evaluating the effects of task design on unfamiliar Francophone listener and automatic speaker identification performance

VoiceCloak

Faced with the threat of identity leakage during voice data publishing, users are engaged in a privacy-utility dilemma when enjoying the utility of voice services. Existing machine-centric studies employ direct modification or text-based re-synthesis to de-identify users' voices but cause inconsistent audibility for human participants in emerging online communication scenarios, such as virtual meetings. In this paper, we propose a human-centric voice de-identification system, VoiceCloak, which uses adversarial examples to balance the privacy and utility of voice services. Instead of typical additive examples inducing perceivable distortions, we design a novel convolutional adversarial example that modulates perturbations into real-world room impulse responses. Benefiting from this, VoiceCloak could preserve user identity from exposure by Automatic Speaker Identification (ASI), while remaining the voice perceptual quality for non-intrusive de-identification. Moreover, VoiceCloak learns a compact speaker distribution through a conditional variational auto-encoder to synthesize diverse targets on demand. Guided by these pseudo targets, VoiceCloak constructs adversarial examples in an input-specific manner, enabling any-to-any identity transformation for robust de-identification. Experimental results show that VoiceCloak could achieve over 92% and 84% successful de-identification on mainstream ASIs and commercial systems with excellent voiceprint consistency, speech integrity, and audio quality.

Closed-set automatic speaker identification using multi-scale recurrent networks in non-native children.

Children may benefit from automatic speaker identification in a variety of applications, including child security, safety, and education. The key focus of this study is to develop a closed-set child speaker identification system for non-native speakers of English in both text-dependent and text-independent speech tasks in order to track how the speaker's fluency affects the system. The multi-scale wavelet scattering transform is used to compensate for concerns like the loss of high-frequency information caused by the most widely used mel frequency cepstral coefficients feature extractor. The proposed large-scale speaker identification system succeeds well by employing wavelet scattered Bi-LSTM. While this procedure is used to identify non-native children in multiple classes, average values of accuracy, precision, recall, and F-measure are being used to assess the performance of the model in text-independent and text-dependent tasks, which outperforms the existing models.

Speaker identification and localization using shuffled MFCC features and deep learning

The use of machine learning in automatic speaker identification and localization systems has recently seen significant advances. However, this progress comes at the cost of using complex models, computations, and increasing the number of microphone arrays and training data. Therefore, in this work, we propose a new end-to-end identification and localization model based on a simple fully connected deep neural network (FC-DNN) and just two input microphones. This model can jointly or separately localize and identify an active speaker with high accuracy in single and multi-speaker scenarios by exploiting a new data augmentation approach. In this regard, we propose using a novel Mel Frequency Cepstral Coefficients (MFCC) based feature called Shuffled MFCC (SHMFCC) and its variant Difference Shuffled MFCC (DSHMFCC). In order to test our approach, we analyzed the performance of the identification and localization proposed model on the new features at different noise and reverberation conditions for single and multi-speaker scenarios. The results show that our approach achieves high accuracy in these scenarios, outperforms the baseline and conventional methods, and achieves robustness even with small-sized training data.

An Efficient Text-Independent Speaker Identification Using Feature Fusion and Transformer Model

Automatic Speaker Identification (ASI) involves the process of distinguishing an audio stream associated with numerous speakers’ utterances. Some common aspects, such as the framework difference, overlapping of different sound events, and the presence of various sound sources during recording, make the ASI task much more complicated and complex. This research proposes a deep learning model to improve the accuracy of the ASI system and reduce the model training time under limited computation resources. In this research, the performance of the transformer model is investigated. Seven audio features, chromagram, Mel-spectrogram, tonnetz, Mel-Frequency Cepstral Coefficients (MFCCs), delta MFCCs, delta-delta MFCCs and spectral contrast, are extracted from the ELSDSR, CSTR-VCTK, and Ar-DAD, datasets. The evaluation of various experiments demonstrates that the best performance was achieved by the proposed transformer model using seven audio features on all datasets. For ELSDSR, CSTR-VCTK, and Ar-DAD, the highest attained accuracies are 0.99, 0.97, and 0.99, respectively. The experimental results reveal that the proposed technique can achieve the best performance for ASI problems.

A hybrid of Deep Neural Network and eXtreme Gradient Boosting for Automatic Speaker Identification

A hybrid of Deep Neural Network and eXtreme Gradient Boosting for Automatic Speaker Identification

Speaker Naming in Arabic TV Programs

Automatic speaker identification is the problem of identifying speakers by their real identities. Previous approaches use textual information as a source of naming, try to associate names to neighbouring speaker segments using linguistic rules. However, these approaches have a few limitations that hinder their application on spoken text. Deep learning approaches for natural language processing have recently reached state-of-the-art results. However, deep learning requires a lot of annotated data which is difficult to obtain in the case of speaker identification task. In this paper, we present two contributions towards integrating deep learning for identifying speakers in news broadcasts: first we realise a dataset in which the names of mentioned speakers are related to the previous, next, current or other speaker turns. Moreover, we present our approach to solve the problem of speaker identification using information obtained from the transcription. We use a Long-term Recurrent Convolutional Network for name assignment and integer linear programming for name propagation into the different segments. We evaluate our model on both assignment and propagation tasks on the test part of the Arabic multi-genre broadcast dataset which consists of 17 TV programs from Aljazeera. The performance is analysed using the evaluation metrics, such as Estimated Global Error Rate (EGER) and Diarization Error Rate (DER). The outcome of the proposed method ensures better performance by achieving the lower EGER of 32.3% and DER of 8.3%.

A Comparative Assessment of Text-independent Automatic Speaker Identification Methods Using Limited Data

Automatic Speaker Identification (ASI) is one of the active fields of research in signal processing. Various machine learning algorithms have been used for this purpose. With the recent developments in hardware technologies and data accumulation, Deep Learning (DL) methods have become the new state-of-the-art approach in several classification and identification tasks. In this paper, we evaluate the performance of traditional methods such as Gaussian Mixture Model-Universal Background Model (GMM-UBM) and DL-based techniques such as Factorized Time-Delay Neural Network (FTDNN) and Convolutional Neural Networks (CNN) for text-independent closed-set automatic speaker identification on two datasets with different conditions. LibriSpeech is one of the experimental datasets, which consists of clean audio signals from audiobooks, collected from a large number of speakers. The other dataset was collected and prepared by us, which has rather limited speech data with low signal-to-noise-ratio from real-life conversations of customers with the agents in a call center. The duration of the speech signals in the query phase is an important factor affecting the performances of ASI methods. In this work, a CNN architecture is proposed for automatic speaker identification from short speech segments. The architecture design aims at capturing the temporal nature of speech signal in an optimum convolutional neural network with low number of parameters compared to the well-known CNN architectures. We show that the proposed CNN-based algorithm performs better on the large and clean dataset, whereas on the other dataset with limited amount of data, traditional method outperforms all DL approaches. The achieved top-1 accuracy by the proposed model is 99.5% on 1-second voice instances from LibriSpeech dataset.

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.

Simultaneous speaker identification and watermarking

Biometric template protection of speech signals and information hiding in speech signals are two challenging issues. To resolve such limitations and increase the level of security, our objective is to build multi-level security systems based on speech signals. So, speech watermarking is used simultaneously with automatic speaker identification. The speech watermarking is performed to embed images into the speech signals that are used for speaker identification. The watermark is extracted for authentication, and then the effect of watermark removal on the performance of the speaker identification system in the presence of degradations is studied. This paper presents an approach for speech watermarking based on empirical mode decomposition (EMD) in different transform domains and singular value decomposition (SVD). The speech signal is decomposed in different transform domains with EMD to yield zero-mean components called intrinsic mode functions (IMFs). The watermark is inserted into one of these IMF components with SVD. A comparison between different transform domains for implementing the proposed watermarking scheme on different IMFs is presented. The log-likelihood ratio (LLR), correlation coefficient (Cr), signal-to-noise ratio (SNR), and spectral distortion (SD) are used as metrics for the comparison. According to the simulation results, we find that the watermark embedding in the discrete sine transform domain provides higher SNR and Cr values and lower SD and LLR values. The proposed approach is robust to different attacks.

Arabic digits speech recognition and speaker identification in noisy environment using a hybrid model of VQ and GMM

This paper presents an automatic speaker identification and speech recognition for Arabic digits in noisy environment. In this work, the proposed system is able to identify the speaker after saving his voice in the database and adding noise. The mel frequency cepstral coefficients (MFCC) is the best approach used in building a program in the Matlab platform; also, the quantization is used for generating the codebooks. The Gaussian mixture modelling (GMM) algorithms are used to generate template, feature-matching purpose. In this paper, we have proposed a system based on MFCC-GMM and MFCC-VQ Approaches on the one hand and by using the Hybrid Approach MFCC-VQ-GMM on the other hand for speaker modeling. The White Gaussian noise is added to the clean speech at several signal-to-noise ratio (SNR) levels to test the system in a noisy environment. The proposed system gives good results in recognition rate.

Un primer acercamiento a la caracterización acústica del habla de niños costarricenses

A medida que la interacción de las personas con las computadoras se hace más extendida, se vuelve más importante el desarrollo de tecnologías para el reconocimiento automático de la voz, la síntesis de voz, así como otras tecnologías relacionadas, considerando a personas de todas las edades, acentos y condiciones. Uno de los grupos humanos que representan desafíos más grandes es el de los niños, debido a las dificultades para grabar suficientes recursos de habla, y la falta de caracterización de su forma de hablar, la cual es particular de cada idioma y acento. Este artículo presenta una primera aproximación para el análisis acústico de niños costarricenses de entre seis y doce años. Estos análisis tienen como objetivo lograr una mejor comprensión de las características del habla producida por este grupo en particular, en términos de propiciar el desarrollo y mejora de los reconocedores automáticos del habla y los sistemas de identificación del hablante.
 Para este propósito realizamos grabaciones de habla de tres niños, las cuales consistieron en palabras aisladas, y comparamos los resultados con tres adultos en términos de los formantes de las vocales. Los formantes proporcionan información sobre el tracto vocal del hablante, y es un parámetro importante para establecer un primer análisis de estas señales. Los resultados muestran diferencias notables entre los niños y los adultos y pueden brindar información útil para futuros estudios en términos de adaptar y desarrollar las tecnologías del habla para esta población.

Speaker Identification in Multi-Talker Overlapping Speech Using Neural Networks

Although numerous works have studied the problem of automatic speaker identification (SID), there are only few works on the SID for overlapping speech, and none of them consider the case of more than two simultaneous speakers. Recognizing that overlapping speech occurs frequently in real-life scenarios, such as in meetings or debates, this work investigates the methods for overlapping SID (OSID) that can determine identities in the overlapping speech from up to five simultaneous speakers. We propose two deep-learning OSID systems, one is two-stage and the other is single-stage. The two-stage system determines the number of simultaneous speakers firstly, followed by identifying the speaker(s). The single-stage system uses a single classifier to perform OSID directly, which is slightly more computationally efficient than the two-stage system. Our experiments show that the two-stage OSID system achieves better identification accuracy than that of the single-stage system. In addition, both the OSID systems based on one-dimensional convolutional neural networks (1DCNN) perform better than the systems based on multilayer perceptron (MLP) and Gaussian mixture models (GMMs). The proposed 1DCNN-based two-stage OSID system achieves 98.55% OSID accuracy for the clean audio data containing up to five simultaneous speakers. In more challenging experimental conditions involving both background noises and high overlapping energy ratios, the system still attained accuracies of above 90%.

Automatic Speaker Identification by Voice Based on Vector Quantization Method

In this paper, the systems of speaker identification of a text-dependent and independent nature were considered. Feature extraction was performed using chalk-frequency cepstral coefficients (MFCC). The vector quantization method for the automatic identification of a person by voice has been investigated. Using the extracted features, the code book from each speaker was built by clustering the feature vectors. Speakers were modeled using vector quantization (VQ). Using the extracted features, the code book from each speaker was built by clustering the feature vectors. Codebooks of all announcers were collected in the database. From the results, it can be said that vector quantization using cepstral features produces good results for creating a voice recognition system.