Abstract

CHANCHI-GOLONDRINO, GABRIEL ELÍAS; RODRIGUEZ-BACA, LISET S.  and  SIERRA-MARTINEZ, LUZ MARINA. Speech Emotion Recognition Software System for Forensic Analysis. Ing. Desarro. [online]. 2024, vol.42, n.1, pp.1-.  Epub Apr 17, 2024. ISSN 0122-3461.  https://doi.org/10.14482/inde.42.01.519.019.

Affective computing aims to create systems capable of recognizing, processing, and simulating human emotions to enhance human-computer interaction. Speech emotion recognition (SER) is a highly effective and non-invasive technique for assessing a user's emotions by analyzing physiological variables. However, despite its widespread use in end-user perception identification, few applications have been developed in the field of forensic analysis. To address this gap, this research proposes a new forensic emotion analysis software system, FOREMAN, based on the emotional study of the voice. The system was developed using the Iterative Research Pattern. FOREMAN determines the fluctuation of emotions in an audio file, as well as the percentage and spatial distribution of emotions present in its segments, using clustering methods. The system's effectiveness is demonstrated by its application to an audio file taken from a Colombian Special Jurisdiction for Peace (JEP) hearing.

Keywords : affective computing; arousal; emotion models; forensic; interviews; valence.

        · abstract in Spanish     · text in English     · English ( pdf )