Rouser is a machine listening system built to analyse and extract information from key live audio sources in order to predict ‘excitement’ events , for example when a player is likely to score a try.
Analog Folk asked UV Tech to find a sonic solution for an application which was launched as part of the Rugby World Cup campaign. The project involved the analysis and extraction of three key audio sources – network TV commentary, crowd audio and on-field audio signals to predict events of significance.
Once detected, the audience was notified via the app that something of interest was ‘likely’ to happen so that they could shift their attention to watch the game.
The realtime machine listening signals were routed through a simple neural net which set off a chain of thresholds ultimately leading to an event detection.
Client: Analog Folk
Head of Music and Innovation: Charlton Hill
Creative Direction: Justin Shave
Coding: Justin Shave/Matt Perrot
Studio: Uncanny Valley