PENGENALAN SUARA PEMBICARA BERDASARKAN SINYAL SUARA BERBAHASA INDONESIA UNTUK MENDUKUNG AKUSTIK FORENSIK

Abstract

Nowadays, crime rate in Indonesia are increasing. People attempt crimes in many different various ways. One of the most frequent acts we met is by changing appearance, and changing their voice to trick the target. Therefore it is very important for us to know who we are talking to. Different emotions can be an obstacle for us to recognize the voice of the other person. This research is conducted to learn more about voiceprint recognition by processing emotional speech signals in Indonesian language using Mel Frequency Cepstral Coefficient (MFCC) for its features and Support Vector Machine (SVM) to classify the features. The result suggests that the recognition can achieve about 92% of the level of accuracy.