Jeremy Christian Undergraduate Dissertation 2015/16

A Platform for Processing Multimedia Streams from an Instrumented Meeting Room

Supervised by T.Hain

Abstract

Media recordings using multiple instruments are becoming more commonplace, however linking the instruments to a unified framework is often difficult. This report details the creation of a media framework designed to track speakers in a meeting environment, recording what they say and do through both video and audio media before transcribing their speech and presenting the user with a subtitled video of their meeting. The framework was unable to implement beamforming algorithms to aid the audio processing element of the project, however it was able to follow users using a motorized camera and record their speech for use in an automated speech recognition service, WebASR.