Recent approaches towards Speaker Anonymization

Abstract

Voice interfaces are an increasingly popular means of interaction with smart objects and applications. The underlying technologies must be trained on large amounts of speech and text data. This approach (i) raises critical privacy concerns, (ii) fails to address languages and categories of users for which little data is available, and (iii) incurs high costs for voice technology companies and application developers. The overall objective of COMPRISE is to define a fully private-by-design methodology that will reduce the cost and increase the inclusiveness of voice interaction technology through research advances on privacy-driven data transformation, on-the-fly translation, user personalisation, and automated data annotation. The resulting software tools will be integrated in an easy-to-use Software Development Kit (SDK) interoperating with a Cloud Platform.