In this paper we propose novel, audio-visual analysis techniques for event detection in broadcast TV sports video content. The scope of the design is constrained to the specialized domain of 'field sport', and specifically, Gaelic Football is presented as an experimental case study. We show that a combination of speech-band energy tracking in the audio domain, coupled with colour dominance pattern recognition in the video domain, provides a useful contribution to event detection for broadcast Gaelic Football matches. It is projected that, any conclusions made therein may be extended such that they function on sports content of a similar nature such as American Football, Australian Rules, Rugby Union etc.