Speaker Recognition for Device Controlling using MFCC and GMM Algorithm

Biometric technology is widely used to identify a smart home device controller with access control to the system. Sound Abstract is one of the biometric technologies used because human speech is different and unique. Generally, a smart home device controller based on sound can be controlled by everyone so that a speaker who should not have access rights to the system will still execute his voice command. The solution to this problem is a sound control system that can identify one speaker's voice with other speakers registered on the system to control smart home devices and reject commands from foreign speakers who are not registered on the system to secure a voice control system is formed. The Mel-Frequency Cepstrum Coefficient (MFCC) method, capable of capturing the characteristics of different human voices and is unique; the output of the MFCC is modeled and classified using GMM (Gaussian Mixture Model) on each cepstrum subject so that the modeling results can identify the voice of the speaker registered on the system listed or the voice of foreign speakers not registered with the system. The accuracy of the system built can identify the voice of the speaker registered on the system by 98.1% and reject the voice of the speaker who is not registered on the system by 91.6%.