IMPLEMENTATION OF FORENSIC METHOD USING VOICE RECOGNITION TECHNIQUE TO ANALYZE VOICE RESEMBLANCE TOWARDS MOBILE PHONE’S VOICE RECORDER

Abstract

This article is intended to analyze the implementation of the forensic method using voice ecognition 
technology to analyze voice resemblance towards mobile phone's voice recorder, by using Forensic audio  method In Audio forensics: Theory and Analysis, namely Pitch Statistical Analysis, Forman and Bandwidth  Statistical Analysis, Graphical Distribution Analysis and Spectrogram Analysis. However, in this study,  researchers only focus on identifying pitch and formants in the data to be analyzed. This study provides an  example scenario of applying digital forensic techniques for voice pain recognition. Greetings compare the  sound in the evidence and the sound suspect on mobile phone media with a voice case study woman. In this  study, PRAAT application was used to help with the audio comparison process from Known Samples and  Unknown Samples. The results of each sample of digital evidence of native speakers' voices and recorded  comparisons on the two assessment factors, namely Pitch and Formant, the researcher cannot distinguish  between the original speaker's voice and the comparison voice. So it is hoped that the output of this research  can become a reference or enrichment material for law enforcement agencies, law, and academics who wish to  continue their research related to digital forensics. The researcher can state that the accuracy of the pitch and  formant analysis through PRAAT is very low and cannot be used as an indicator to distinguish that a date is  correct as of the original voice, as evidenced by the results of the F0 value of the original voice data and the  comparison obtained. Very flexible and inconsistent. Some are the same, close to and even far from the F0 value of the subject data. This shows that everyone has a different pitch value because each person's word  pronunciation is different. There may be pitch values from several subjects that are almost the same.