Echo Cancellation in the Access Grid

Exhibitor: David Hunter

Supervisor: Geoff Ericksson

Research Group: Complex and Intelligent Systems

Industry Sector: Information and Communication Technology (ICT)

Acoustic echoes can arise in many systems that involve remote audio transmission, causing distracting feedback and disrupting communication. In the Access Grid, a high-quality Internet-based video conferencing network, existing solutions to remove these echoes are either expensive or inadequate. In response to this situation, a software system was developed to address the challenge of removing acoustic echoes in real time, with the aim of enabling low-cost acoustic echo cancellation in an Access Grid node.

The system was implemented to integrate with the existing Access Grid audio tool, and was designed to detect and cancel locally-generated echoes through echo path modelling using an adaptive algorithm. While the adaptive algorithm used in the solution did not provide a sufficiently accurate model of the echo environment to successfully remove echoes, the system developed provided a well-structured, extensible base for performing real-time software echo cancellation in the Access Grid.

Thesis Document (PDF)