Smart Speaker Privacy Control - Acoustic Tagging for Personal Voice Assistants

Cheng, Peng and Bagci, Ibrahim Ethem and Yan, Jeff and Roedig, Utz (2019) Smart Speaker Privacy Control - Acoustic Tagging for Personal Voice Assistants. In: SafeThings 2019 : IEEE Workshop on the Internet of Safe Things. IEEE.

Text (Smart Speaker Privacy Control - Acoustic Taggingfor Personal Voice Assistants)
Acoustic_tagging_and_watermarking_2.pdf - Other
Download (10MB)

Abstract

Personal Voice Assistants (PVAs) such as the Siri, Amazon Echo and Google Home are now commonplace. PVAs continuously monitor conversations which may be transported to a cloud back end where they are stored, processed and maybe even passed on to other service providers. At present, a user has little control over this process. He is unable to control the recording behaviour of surrounding PVAs, is unable to signal his privacy requirements to back-end systems and is unable to track conversation recordings. In this paper we explore techniques for embedding additional information into acoustic signals processed by PVAs. A user employs a tagging device which emits an acoustic signal when PVA activity is assumed. Any active PVA will embed this tag in the recorded audio stream. The tag may signal a cooperating PVA or back-end system that a user has not given a recording consent. The tag may also be used to trace when and where a recording was taken. In this paper we discuss different tagging techniques and application scenarios. We describe the implementation of a prototype tagging device based on PocketSphinx. Using the popular PVA Google Home Mini we demonstrate that the device can tag conversations and that the tagging signal can be retrieved from conversations stored in the Google back-end system.