A proposal for a tool for experimenting with deep learning in speech recognition for the Portuguese language

Main Article Content

Julio Cesar Duarte
Matheus Rodrigues Affonso
Rebeca Pinheiro dos Reis

Abstract

Natural Language Processing is now part of the routine of a large part of society, in simple gestures such as voice commands mobile phones or spoken instructions to a virtual assistant. Understanding what is said verbally can be a simple task for a native speaker of a certain language, and it is considered one of the most essential for interpersonal interaction. In the current context, marked by the great search for integration between systems, Internet of Things and environments equipped with intelligent devices and robots, such interaction between men and machines is in evidence. An approach using machine learning to act in this relationship becomes a possible solution. The present work aims at presenting an environment that allows the experimentation of models for the Speech Recognition task as well as implementing classification models using deep learning neural networks. At the end, a comparative analysis between the accuracy of possible models being tested is made. As contributions of this work, a methodology that allows the experimentation of speech recognition and the proposal of a trained classifier using a Brazilian Portuguese Audio dataset are presented.

Downloads

Download data is not yet available.

Article Details

How to Cite
Duarte, J. C., Affonso, M. R., & Reis, R. P. dos. (2022). A proposal for a tool for experimenting with deep learning in speech recognition for the Portuguese language. Revista Militar De Ciência E Tecnologia, 38(4), 83-95. Retrieved from http://ebrevistas.eb.mil.br/CT/article/view/9047
Section
Artigos

Most read articles by the same author(s)