Implementation of Audio Forensic Methods in the Identification of Digital Evidence from Voice Recordings

Abstract

The advancement of technology has increased convenience in various daily activities, but it has also heightened the potential for misuse in criminal acts such as fraud, corruption, theft, and the dissemination of hoax information. Digital evidence, including files, images, audio, and video, often plays a crucial role in criminal investigations. Each type of digital evidence requires a specific analytical method, one of which is audio forensics. Audio forensics is a sound analysis method aimed at enhancing speech clarity and reducing noise, thereby providing important clues in investigative processes. Tools such as Praat are used to compare allegedly manipulated voice recordings with original voice samples. This study aims to identify digital evidence in the form of voice recordings using Praat and audio forensic methods. By analyzing voice characteristics such as pitch, formant, and spectrogram, this research seeks to determine the similarity between evidentiary recordings and reference samples as well as to perform audio enhancement. The results indicate that in the pitch analysis stage, the samples of Yuki, Sakinah, Sonia, Marwah, and Salsa were identical to the evidence recordings. Furthermore, formant analysis revealed that the samples of Yuki, Sonia, and Salsa were identical to the evidence, while spectrogram analysis showed that only the Salsa sample matched the evidence recordings.