To search, Click below search items.


All Published Papers Search Service


People Recognition through Footstep Sound Using MFCC Extraction Method of Artificial Neural Network Back Propagation


Jan Everhard Riwurohi, Jazi Eko Istiyanto, Khabib Mustofa, Agfianto Eko Putra


Vol. 18  No. 4  pp. 28-35


The sound of footsteps is a sound that is heard when people are walking, where the sound is the result of a jerk between the foot and the floor surface. This sound may vary from individual to individual due to some differences, including how to step, footwear used, the characteristics of the floor, height, and weight of the person. The sound of this footstep is a biometric feature and can be used as a person's identifier when the person is stepping. In some previous studies, the process of recording the footsteps on the data acquisition media is done on one trackside only, which is on the middle side of the track, as they used only one microphone. It means that the footsteps sound used to recognize a person was only the closest footsteps sound from the microphone. In this study, the sound of footsteps is recorded at the time when people are walking not only on one side of the course but three sides of different paths - the left side of the track, the middle side of the track and on the right side of the track. Thus, the footsteps sound used to recognize a person all the footsteps sounds recorded from all of those sides of the track. This way is aimed to the process of recognition of people based on the sound of footsteps on a data acquisition medium can run as natural as possible so that wherever the person stepped can be recognized well based on the sound of footsteps. This study created a media to support data acquisition process and to support the process of footstep recording and to put the microphone. The study used four microphones which are placed on the right and the left side of data acquisition track. The method of feature extraction used to extract the footstep sound is Mel Frequency Cepstrum Coefficient (MFCC), and the classification method for the recognition process of the person is using Back Propagation Neural Network. The accuracy of person recognition based on footsteps obtained in this study is 98.8% for the recognition of people walking on the left side of the track, 98.8% on the middle side of the track and 95% for the recognition of the person running on the right side of the track.


Footsteps sound, data acquisition media, microphone, MFCC, Artificial Neural Network Back Propagation.