Aurigin AuriginInspect

AuriginInspect is a deepfake detection platform designed for forensic use in legal proceedings and media investigations. Users upload an audio exhibit and receive a layered, explainable analysis with a structured report intended for expert witness testimony. The platform performs calibrated detection using likelihood ratio scoring to distinguish genuine from synthetic audio, with characterized uncertainty. It provides segment localisation, offering time-stamped, region-by-region classification to identify where synthetic content appears within a recording. Planned capabilities include acoustic attribution (ranking cues such as prosody anomalies, spectral artifacts, noise inconsistencies, and breathing patterns), model transparency (displaying model version, benchmark dataset, and metrics such as EER and Cllr), and generation of a forensic PDF report containing metadata, waveform annotations, attribution, calibration, and methodology. The product is positioned to support ISO 27042-aligned, documented and reproducible methodology with an audit trail from upload to report, and to maintain chain of custody through file hash, upload timestamp, model version, and analysis parameter logging. Deployment options include a secure web application requiring no installation, API integration to embed detection into existing forensic software, compliance systems, and media verification pipelines, and an on-premises containerised deployment for forensic labs where evidence cannot traverse external networks. The service is Swiss-hosted and stated to be compliant with nFADP and GDPR.