Leviathan’s Ear: Tracing Machine Listening in Android Applications

The team

• Dr Iain EmsleyLink opens in a new window (Project Lead), Warwick Manufacturing Group

The project

Leviathan’s Ear is a project to engage with the political economy of machine listening. Voice is a personal and changing entity that use a variety of signals, such as register, speech, or spatial position, as well as encoded semantic information. Yet, machine listening processes, such as transcription, transform the voice into signals for processing. This project aimed to consider the ways in which voice had become an interface to machine learning and how such services are embedded into everyday life.

The project had three stages. Firstly, Google’s public page was scraped for their voice models and data centres to understand their limits and the availability of voice and models in reginal data centres. The second stage is a software reading of the public code provided by Google and Android to use Google’s Speech to Text infrastructure. These stages were to be presented as posters at the workshop.

The final stage was a workshop that was streamed via YouTube to interview the machinelisteningLink opens in a new window.

Project outcomes

The project held a workshop with fellow academics in Australia that was live-streamedLink opens in a new window. The software reading is still underway and will be extended from a poster to short article. 

We have had 91 streams of the YouTube video after the stream and 10 viewers while it happened. We had 5 attendees from the Centre for Interdisciplinary Methodologies, English, and the Department of Music, Oxford University.

Enhancing interdisciplinarity

An AHRC bid is being prepared from some of the strands identified.