Security System Using A Robot Based On Speech Recognition

This study describes a security system using a humanoid robot by utilizing speech recognition. The robot has two main parts, namely, Raspberry Pi 3 and two Arduino UNO R3 as a slave. This robot is designed as a combination of speech recognition and voice biometric. The instruction given by a speaker must be obeyed by the robot using servo motor. Meanwhile, for voice biometric, robot may give access to an authorized person using speech recognition. Mel Frequency Cepstral Coefficients (MFCCs), their delta, and delta-delta are used as feature extraction which is fed to a classifier, Gaussian Mixture Model (GMM). Results of this study show that the robot may recognize the speaker with an accuracy of 99.4% and 99% for 50% of testing data and 20% of testing data, respectively. Thus, this suggests that the combination of MFCC and GMM can be implemented in speech recognition for security system performed by the robot.