Voice verification using i-vectors and neural networks with limited training data

This study proposes an approach to voice identification based on neural networks (DNN) for i-Vector. Modern voice identification systems based on DNN use large amounts of labeled training data. Using the LRE i-Vector Machine Learning Challenge restricts access to ready-to-use i-Vector for learning...

Full description

Bibliographic Details
Main Authors: Mamyrbayev, Orken Zh., Othman, Mohamed, Akhmediyarova, A. T., Kydyrbekova, Aizada S., Mekebayev, Nurbapa O.
Format: Article
Language:English
Published: National Academy of Sciences of the Republic of Kazakhstan 2019
Online Access:http://psasir.upm.edu.my/id/eprint/82733/
http://psasir.upm.edu.my/id/eprint/82733/1/Voice%20verification%20.pdf
_version_ 1848859340158533632
author Mamyrbayev, Orken Zh.
Othman, Mohamed
Akhmediyarova, A. T.
Kydyrbekova, Aizada S.
Mekebayev, Nurbapa O.
author_facet Mamyrbayev, Orken Zh.
Othman, Mohamed
Akhmediyarova, A. T.
Kydyrbekova, Aizada S.
Mekebayev, Nurbapa O.
author_sort Mamyrbayev, Orken Zh.
building UPM Institutional Repository
collection Online Access
description This study proposes an approach to voice identification based on neural networks (DNN) for i-Vector. Modern voice identification systems based on DNN use large amounts of labeled training data. Using the LRE i-Vector Machine Learning Challenge restricts access to ready-to-use i-Vector for learning and testing the voice identification system. This poses unique challenges in developing DNN-based voice identification systems, since optimized external interfaces and network architectures can no longer be used. We propose to use the training i-Vectors to train the initial DNN to identify the voice. Next, we present a novel strategy for using this initial DNN to strip the language labels of the inappropriate set from the development data. The final DNN for voice identification is trained using the original training data and the estimated out-of-set language data. We show that augmenting the training set with out-of- set labels leads to a significant improvement in voice identification performance. In this paper, we studied the possibility of using neural networks for speech identification. In particular, standard approaches to speech recognition were considered, the concept of an artificial neuron as an object used in speech identification was defined. A speech recognition option using a neural network was investigated, and steps were presented to perform this task. Accuracy using neural networks with limited learning data and a higher i-vector dimension is superior to others with a score of 92.1%. From this study, we can conclude that the size of the UBM and the dimension of the i-vector affect the accuracy of voice identification based on the i-vector.
first_indexed 2025-11-15T12:27:47Z
format Article
id upm-82733
institution Universiti Putra Malaysia
institution_category Local University
language English
last_indexed 2025-11-15T12:27:47Z
publishDate 2019
publisher National Academy of Sciences of the Republic of Kazakhstan
recordtype eprints
repository_type Digital Repository
spelling upm-827332021-06-05T00:12:57Z http://psasir.upm.edu.my/id/eprint/82733/ Voice verification using i-vectors and neural networks with limited training data Mamyrbayev, Orken Zh. Othman, Mohamed Akhmediyarova, A. T. Kydyrbekova, Aizada S. Mekebayev, Nurbapa O. This study proposes an approach to voice identification based on neural networks (DNN) for i-Vector. Modern voice identification systems based on DNN use large amounts of labeled training data. Using the LRE i-Vector Machine Learning Challenge restricts access to ready-to-use i-Vector for learning and testing the voice identification system. This poses unique challenges in developing DNN-based voice identification systems, since optimized external interfaces and network architectures can no longer be used. We propose to use the training i-Vectors to train the initial DNN to identify the voice. Next, we present a novel strategy for using this initial DNN to strip the language labels of the inappropriate set from the development data. The final DNN for voice identification is trained using the original training data and the estimated out-of-set language data. We show that augmenting the training set with out-of- set labels leads to a significant improvement in voice identification performance. In this paper, we studied the possibility of using neural networks for speech identification. In particular, standard approaches to speech recognition were considered, the concept of an artificial neuron as an object used in speech identification was defined. A speech recognition option using a neural network was investigated, and steps were presented to perform this task. Accuracy using neural networks with limited learning data and a higher i-vector dimension is superior to others with a score of 92.1%. From this study, we can conclude that the size of the UBM and the dimension of the i-vector affect the accuracy of voice identification based on the i-vector. National Academy of Sciences of the Republic of Kazakhstan 2019-06 Article PeerReviewed text en http://psasir.upm.edu.my/id/eprint/82733/1/Voice%20verification%20.pdf Mamyrbayev, Orken Zh. and Othman, Mohamed and Akhmediyarova, A. T. and Kydyrbekova, Aizada S. and Mekebayev, Nurbapa O. (2019) Voice verification using i-vectors and neural networks with limited training data. Bulletin of the National Academy of Sciences of the Republic of Kazakhstan, 3 (379). pp. 36-43. ISSN 1991-3494; ESSN: 2518-1467 https://www.researchgate.net/publication/333891112_VOICE_VERIFICATION_USING_I-VECTORS_AND_NEURAL_NETWORKS_WITH_LIMITED_TRAINING_DATA 10.32014/2019.2518-1467.66
spellingShingle Mamyrbayev, Orken Zh.
Othman, Mohamed
Akhmediyarova, A. T.
Kydyrbekova, Aizada S.
Mekebayev, Nurbapa O.
Voice verification using i-vectors and neural networks with limited training data
title Voice verification using i-vectors and neural networks with limited training data
title_full Voice verification using i-vectors and neural networks with limited training data
title_fullStr Voice verification using i-vectors and neural networks with limited training data
title_full_unstemmed Voice verification using i-vectors and neural networks with limited training data
title_short Voice verification using i-vectors and neural networks with limited training data
title_sort voice verification using i-vectors and neural networks with limited training data
url http://psasir.upm.edu.my/id/eprint/82733/
http://psasir.upm.edu.my/id/eprint/82733/
http://psasir.upm.edu.my/id/eprint/82733/
http://psasir.upm.edu.my/id/eprint/82733/1/Voice%20verification%20.pdf