Verfahren, Computervorrichtung und Computerprogramme zur Verbesserung der Privatsphäre und Sicherheit in der Audio- und Videokommunikation

Details

Veröffentlichungs-Nr.

EP4822689

Anmeldetag

10. April 2025

Veröffentlichung

14. Oktober 2026

Rechtsraum

EP

Abstract

A method, computer device and computer program for enhancing privacy and security in audio and video communications are proposed. The method detects whether an incoming audio stream contains patterns and/or anomalies suggestive of artificially generated synthetic voice by performing a real-time voice analysis including chunking the incoming audio stream into audio segments; identifying audio features by processing the audio segments; producing a sequence of probabilities indicative of a likelihood that the incoming audio stream comprises synthetic voice by a machine learning algorithm processing the identified audio features; computing a probability score based on the produced sequence of probabilities; and determining that the incoming audio stream contains artificially generated synthetic voice if the computed probability score is similar to a given threshold; triggering an alert upon detection of an artificially generated synthetic voice; and anonymizing an outgoing audio stream of the user by modifying or removing user voice signals comprising personal identifiable information.

Anmelder

Firma

Telefonica Innovacion Digital SLU

Land

🇪🇸 Spanien