Phonexia launches voice biometrics product for forensic investigations
Voice biometrics company Phonexia has a new product for forensic voice comparison.
The Phonexia Voice Inspector 5.1 is designed specifically for forensic experts. The Czech company says that the software can identify individuals from just three seconds of conversational speech and offers voice comparison accuracy regardless of language or spoken words. The new software offering meets the international standard for court admissibility, aligning with guidelines from the European Network of Forensic Science Institutes (ENFSI), according to the announcement.
During a forensic test in 2021, using the forensic_eval_01 methodology, the technology achieved an Equal Error Rate (EER) of 2.1 percent before calibration and 1.2 percent EER after calibration. Since then, Phonexia’s team has improved the technology further, the company says in a release.
The product also contains a suite of supportive technologies such as speaker diarization for labeling and separating individual speakers in mono audio recordings based on their voice, a phoneme recognizer for identifying similar sound patterns across recordings, voice activity detection and a spectrogram for in-depth analysis of audio files.
Phonexia is a voice biometrics provider in the EU-backed Roxanne consortium focusing on providing voice biometrics for law enforcement agencies to use in investigating criminal networks. The project was funded by the EU Horizon 2020 innovation program.