This article considers the problem of speech recognition using the example of word recognition learned through the reverse error distribution method.The article presents the practical results of training a constructed neural network with training samples of different sizes.One of the promising areas to solve speech recognition problems is the use of artificial neural networks.Neural networks are widely used to solve various classes of speech recognition problems, taking into account their ability to generalize.A method to solve the speech recognition problem is considered using the example of recognizing individual words of a limited dictionary using a direct propagation neural network trained by the error back propagation method.The practical results of training the constructed neural network at different sizes of the training sample are presented.
| Jurnal | ТАТУ хабарлари |
|---|---|
| Nashr sanasi | 2023-10-03 |
| DOI | 10.61663/233tuitmct1arr |
DOI: 10.61663/233tuitmct1arr · Maqolaning asl sahifasi
This article discusses algorithms and tools for information security through software applications with artificial intelligence and security through devices and technologies.Intelligently designed safe robots help…
The article presents modern methods of cavitation erosion of metal and protection against cavitation, and considers safety issues.The general hypothesis of cavitation erosion, erosion of a material, the occurrence of…
Electricity theft is the biggest problem today, which causes huge losses to electrical panels.In addition, to hide these losses, the price will eventually increase.If we can prevent these thefts, we will be able to save…
Algorithm and program for solving the gas filtration problem in dynamically interconnected porous media
In this article, the limitation of the rate of heat propagation, spatial localization properties of mathematical models of heat propagation and diffusion processes described by the system of nonlinear equations of…
This paper presents a comparative analysis and evaluation of the effectiveness of deep learning-based speech recognition (ASR) models for the Uzbek language. It highlights the growing importance of natural language…
This article proposes the use of a non-invasive method based on measuring the electrical resistance of bioactive points located in the meridians of the human body in diagnosing diabetes.Using analysis of variance, it…
This paper presents a comparative analysis and evaluation of the effectiveness of deep learning-based speech recognition (ASR) models for the Uzbek language. It highlights the growing importance of natural language…
Abstraction:In this article, a mathematical model based on the Hamilton-Ostrogradsky variational principle is presented.Using the Kirkhgoff-Lyav hypothesis, transform the mathematical model in 3D form into his 2D…