Formant-Based Speaker Differentiation in Forensic Audio Analysis: Application, Validation and Development of ForensicSpectralV2.1

ForensicSpectralV2.1 is an open-source forensic tool designed for the visual analysis of formants in speech signals. It enables color-coded, time-synchronous speaker differentiation in forensic audio recordings based on frequency and intensity patterns. The software was developed to address the lack of tools capable of generating court-admissible visual representations of speaker changes in complex recordings. It is particularly useful in cases involving potentially manipulated evidence, overlapped speech, or covert voices. Developed in Python using the well-established Matplotlib library, the tool has been successfully used since 2023 in multiple legal and institutional contexts. These include the detection of audio editing, documentation of LRAD deployments, and visual speaker segmentation in emergency recordings. Version 2.1 will be published on GitHub in May 2025. A commercial version 3.0, with GUI, metadata extraction, and automated segmentation, is planned for the autumn 2025. The source code complies with international standards for audioforensics and aligns with ENFSI and ISO/IEC 27037 recommendations