Introduction to Adversarial Acoustics

This workshop introduces participants to the use of voice in machine learning systems. During this session we will expand upon the arguments made in Not I and introduced participants to some of the methods used.

Through a mix of theoretical discussion and practical computing exercises, we will critically explore the legacy of vocal forensics, tackling the use of the voice as a tool for profiling and measurement as well as exploring the space for an ‟experimental vocal practice” capable of troubling the assumed relationship between speech, subjectivity and the body that AI systems seek to reify.

Participants explore how aesthetic practices can be used as part of an investigative method that opens up computational systems to sociotechnical concerns. In the context of AMRO we will also consider how an investigative practice such as this can adopt the principals of FLOSS.

Following the practical exercises we will facilitate a discussion regarding the ethical problems of reverse engineering and adversarial practices, exploring with participants how potentially dangerous or harmful software can be made accessible to a research community in a responsible manner.

** this workshop is full! registration has been deactivated.