In today’s world, we are surrounded by a variety of Autonomous AUdio Systems (AAUS). Audio capture may contain personal identity attributes of voice and speech that are protected in the scope of individual rights. This data can also be misused by attackers, highlighting key issues of data protection, privacy, and security. AAUS encompasses the wide breadth of devices and systems that are enabled for audio and speech capture, such as home devices, smart watches, virtual assistants, and audio-enabled vehicles.
Prof. Shuo Zhang collaborated with Prof. Jennifer Williams (University of Southampton)’s team in a project in the Trustworthy Autonomous Systems (TAS) Hub under the UKRI in a project “Co-Design of Context-Aware Audio Capture” (link1, link2). In this project, the team co-designs, implements and administers an interactive listening survey wherein participants may choose how to modify audio in various contexts and imagined scenarios where trust is involved. From our analysis, we investigate ways that individual protections and trust of AAUS could be increased.
A new publication this month in the Journal of Computer Speech and Language entitled “Public perceptions of speech technology trust in the United Kingdom” summarized our findings. This is an extended version of our previous publication “Socio-Technical Trust For Multi-Modal Hearing Assistive Technology,” a blue sky paper published in the AMHAT workshop at the 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Greece.