Speech recognition with industrial purposes: an approach using intelligent systems

Resumen

 This paper presents an implementation of Artificial Neural Networks – ANN to recognize voice commands that do not depend on the speaker. From a database having the commands pronounced by different speakers, techniques on voice signal processing were applied in order to have these signals treated within the techniques of Artificial Intelligence, particularly a Neural Network. The definition of which ANN training algorithm to use depends on some factors, such as the complexity of the problem, amount of training data, and precision and accuracy of problems. Multilayer Perceptrons – MLP is a model of Multilayer Neural Network which guarantees good results. Neural Network training uses a powerful voice analysis technique, which is Linear Predictive Coding – LPC. The developed project was considerably empiric, demanding practical tests, graphic and numerical analysis so that the final goal was achieved. The speech recognition, also known as automatic speech recognition or computer speech recognition, converts spoken words into machine-readable input. Speech recognition has great applicability on industry since they can be used to accelerate and optimize trainings and process operation. Beyond that, it can be a useful tool to ensure that people with special needs have more successful job opportunities. Keywords: Speech Recognition, Industrial Processes, Artificial Intelligence, Neural Network, Linear Predictive Coding.