PENGENAL SUARA BERBASIS TEMPLATE-BASED
UNTUK MENGENALI GENDER
Total View This Week0
Total View This Week0
Institusion
INSTITUT TEKNOLOGI DIRGANTARA ADISUTJIPTO
Author
Dimas Nugraha Polusina, Fabiano
Subject
T Technology (General)
Datestamp
2024-03-21 02:14:23
Abstract :
Sound is one of the main human media for communicating in addition to using writing and sign language. Voice recognition system is a system that is responsible for identifying a person from the voice issued when performing commands. The study designed a voice recognition system for recognizing gender using the Fast Fourier Transform (FFT) and Euclidean Distance methods. System design begins with the processing of reference voice data and test sound data to be feasible and easy to classify. Both of these sound data go through the same process, preprocessing and the characteristic extraction process. Then, the results of the extraction of traits using this FFT method will be classified similarly to using the Euclidean Distance. The smallest distance or value of this Euclidean calculation will be determined into the output. Based on the results of system testing, the highest percentage of accuracy achieved by the system is 90% and the lowest is 70%, with an overall average of 82,5%. The results were obtained from tests conducted on 40 test sound samples on 4 reference sound samples from 2 men and 2 women, who said the same word "Hello".
Sound is one of the main human media for communicating in addition to using writing and sign language. Voice recognition system is a system that is responsible for identifying a person from the voice issued when performing commands. The study designed a voice recognition system for recognizing gender using the Fast Fourier Transform (FFT) and Euclidean Distance methods. System design begins with the processing of reference voice data and test sound data to be feasible and easy to classify. Both of these sound data go through the same process, preprocessing and the characteristic extraction process. Then, the results of the extraction of traits using this FFT method will be classified similarly to using the Euclidean Distance. The smallest distance or value of this Euclidean calculation will be determined into the output. Based on the results of system testing, the highest percentage of accuracy achieved by the system is 90% and the lowest is 70%, with an overall average of 82,5%. The results were obtained from tests conducted on 40 test sound samples on 4 reference sound samples from 2 men and 2 women, who said the same word "Hello".
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
