Authors : Shweta Savdekar and Ramchand Hablani
Abstract: This work studies, the application of feedforward Artificial Neural Network (ANN) to address the task of isolated command recognition system for robot control. The study is based on the end-to-end ANN approach for speech recognition system which is interfaced to the Robot Firebird VI for its movement control through voice command. Initially, the database is created by manually recording different instances for each command. For the front end processing of the recorded signals, Mel Frequency Cepstral Coefficient (MFCC) is used as the feature extraction technique. Extracted features are processed by artificial neural network model which works as a classifier. The recognized commands are then communicated to the Firebird VI robot via. wireless connectivity. As a part of the study undertaken, the study also throws light on ways to enhance the performance of the system. The performance of the system is analyzed in terms of the accuracy of the system to correctly recognize the spoken commands and the robot taking the corresponding action.
Shweta Savdekar and Ramchand Hablani, 2017. End-to-End Connectionist Approach for Audio Command Recognition System for Robot Control. Journal of Engineering and Applied Sciences, 12: 8207-8213.