Neural architectures for gender detection and speaker identification

Mamyrbayev, Оrken; Toleu, Alymzhan; Tolegen, Gulmira; Mekebayev, Nurbapa

doi:10.1080/23311916.2020.1727168

Cited by 20 publications

(11 citation statements)

References 20 publications

(18 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For a perfect approximation to the minimum error of the neural network, the learning rate should tend to an infinitesimal value to ensure the best convergence of the learning algorithm. However, the smaller the selected value of the learning step, the longer the learning takes place online [9].…”

Section: Training a Neural Network By Back Propagation Of An Errormentioning

confidence: 99%

Development of the Neural Network for Solving the Problem of Speech Recognition

Dilmagambetova¹,

Mamyrbayev²

2021

News of NAS RK. SPh-M

Self Cite

View full text Add to dashboard Cite

The article discusses a method for solving the problem of speech recognition on the example of recognizing individual words of a limited dictionary using a forward propagation neural network trained by the error back propagation method. The goal was to create a neural network model for recognizing the solution of individual words, analyze the training characteristics and behavior of the constructed neural network. Based on the input data and output requirements, a feedback neural network selected. To train the selected neural network model, a back propagation algorithm was chosen. The developed neural network demonstrated the expected behavior associated with learning and generalization errors. It found that even if the generalization error decreases as the learning sequence increases, the errors begin to fluctuate regardless of the introduction of a dynamic learning rate. The network sufficiently trained to meet the generalization error requirements, but there is stillroom to improve the generalization error. Practical results of training the constructed neural network at different sizes of the training sample presented.

show abstract

Section: Training a Neural Network By Back Propagation Of An Errormentioning

confidence: 99%

Development of the Neural Network for Solving the Problem of Speech Recognition

Dilmagambetova¹,

Mamyrbayev²

2021

News of NAS RK. SPh-M

Self Cite

View full text Add to dashboard Cite

show abstract

“…By trying to simulate objects in the time domain using RNN or convolutional neural networks (CNN) instead of HMM, he encounters the problem of data alignment. The loss functions of both RNN and CNN (Convolutional Neural Networks) are determined at each point in the sequence, therefore, to provide training opportunities, need to know the alignment relationship between the output RNN sequence and the target sequence [13].…”

Section: End-to-end Model Based On Connectionist Temporal Classificationmentioning

confidence: 99%

Thermodynamic Properties of Dense Hydrogen Plasmas With Partially Degenerate Semiclassical Ions

Ismagambetova¹,

Габдуллин²,

Ramazanov³

2021

SPM

View full text Add to dashboard Cite

Б а с р е д а к т о р ы ф.-м.ғ.д., проф., ҚР ҰҒА академигі Ғ.М. Мұтанов Р е д а к ц и я а л қ а с ы: Асанова А.Т. проф. (Қазақстан) Бошкаев К.А.PhD докторы (Қазақстан) Байгунчеков Ж.Ж. проф., академик (Қазақстан) Вишневский И.Н. проф., академик (Украина) Quevedo Hernando проф. (Мексика), Жүсіпов М.А. проф. (Қазақстан) Жұмабаев Д.С. проф. (Қазақстан) Ковалев А.М. проф., академик (Украина) Калимолдаев М.Н. проф., академик (Қазақстан) Михалевич А.А. проф., академик (Белорусь) Молдабеков М. М. проф., академик (Қазақстан) Мырзакулов Р. проф., академик (Қазақстан) Өмірбаев У.У. проф., академик (Қазақстан) Пашаев А. проф., академик (Әзірбайжан) Рамазанов Т.С. проф., академик (Қазақстан) Такибаев Н.Ж. проф., академик (Қазақстан), бас ред. орынбасары Тигиняну И. проф., академик (Молдова) Тулешов А.К. проф., чл.-корр. (Қазақстан) Уалиев З.Г. проф., чл.-корр. (Қазақстан) «ҚР ҰҒА Хабарлары. Физика-математикалық сериясы».

show abstract

“…The neural network used in this work is a fully connected neural network with a straightened linear unit [14] as an activation function. A straightened linear unit is essentially a piecewise function that turns all negative values into 0, while positive values remain unchanged.…”

Section: Fully Connected Neural Network Modelingmentioning

confidence: 99%

Biometric Human Authentication System Through Speech Using Deep Neural Networks (Dnn)

Mamyrbayev¹,

Technologies²,

Akhmediyarova³

et al. 2020

THE NATIONAL ACADEMY OF SCIENCES OF THE REPUBLIC OF KAZAKHSTAN

Self Cite

View full text Add to dashboard Cite

Biometrics offers more security and convenience than traditional methods of identification. Recently, DNN has become a means of a more reliable and efficient authentication scheme. In this work, we compare two modern teaching methods: these two methods are methods based on the Gaussian mixture model (GMM) (denoted by the GMM i-vector) and methods based on deep neural networks (DNN) (denoted as the i-vector DNN). The results show that the DNN system with an i-vector is superior to the GMM system with an i-vector for various durations (from full length to 5s). DNNs have proven to be the most effective features for text-independent speaker verification in recent studies. In this paper, a new scheme is proposed that allows using DNN when checking text using hints in a simple and effective way. Experiments show that the proposed scheme reduces EER by 24.32% compared with the modern method and is evaluated for its reliability using noisy data, as well as data collected in real conditions. In addition, it is shown that the use of DNN instead of GMM for universal background modeling leads to a decrease in EER by 15.7%.

show abstract

Neural architectures for gender detection and speaker identification

Cited by 20 publications

References 20 publications

Development of the Neural Network for Solving the Problem of Speech Recognition

Development of the Neural Network for Solving the Problem of Speech Recognition

Thermodynamic Properties of Dense Hydrogen Plasmas With Partially Degenerate Semiclassical Ions

Biometric Human Authentication System Through Speech Using Deep Neural Networks (Dnn)

Contact Info

Product

Resources

About