Projekt

The consortium investigates both the use and improvement of innovative AI-based anonymization techniques for initial counseling and preventive remote treatment of people who are sexually attracted to children.The Institute of Sexology and Sexual Medicine at the Charité Berlin,here acting as both practical and research partner, has been leading national and international projects for therapy-motivated people with pedophilic or hebephilic inclinations. The goal of AnonymPrevent is speech anonymization that anonymizes a patient's identity (voice, manner of speaking) while retaining the emotion and personality expression content relevant for clinical diagnostic assessment. The trustworthiness of such an AI system is crucial for the utilization of the preventive therapy offer by those seeking help, since participation is associated with shame and fear of social exclusion. The team also investigates whether anonymization of the verbal communication channel leads to more acceptance for preventive treatment against child abuse and promotes an open exchange without unfavorably influencing the communication in the therapy.

• Prof. Dr.-Ing. Ingo Siegert

  Universität Magdeburg
  Fakultät für Elektro- und Informationstechnik
  Institut für Informations- und
  Kommunikationstechnik
  Magdeburg
  

• Prof. Dr.-Ing. Sebastian Stober

  Universität Magdeburg
  Fakultät für Informatik
  Institut für Intelligente Kooperierende System
  Magdeburg
  

• Prof. Dr.-Ing. Sebastian Möller

  Technische Universität Berlin
  Institut für Softwaretechnik und Theoretische
  Informatik
  Quality and Usability Lab
  Berlin
  

• Dr.-Ing. Tim Polzehl

  Technische Universität Berlin
  Quality and Usability
  Institute of Software Engineering and Theoretical
  Berlin
  

• Prof. Dr. Dr. Klaus Beier

  Charité - Universitätsmedizin Berlin
  Centrum für Human- und Gesundheitswissenschaften
  Institut für Sexualwissenschaft und Sexualmedizin
  Berlin
  

• Towards Emotion Preserving Voice Conversion Using Deep Embeddings

• Voice Privacy - leveraging multi-scale blocks with ECAPA-TDNN SE-Res2NeXt extension for speaker anonymization

• Why Eli Roth should not use TTS-Systems for anonymization.