Speech to text transcription is constantly improving in accuracy and becoming more wide-spread. However most speech-to-text systems require close-talking microphones to achieve sufficient audio quality. This is not always practical especially in military installations, where wireless lapel microphones pose a security risk. This talk considers speech-to-text transcription in a video conferencing environment using distant microphones. Several options for improving the audio quality are considered including directional microphones and microphone beamforming techniques. The increase in word recognition accuracy for each approach are presented.
Mr Matthew Trinkle Received the B.Eng (Hons) degree in Electrical Electronic Engineering and B.Sc. in Mathematical and Computer Science from the University of Adelaide, South Australia, in 1994 and 1995 respectively. From 1996 to 2004 he was with the Cooperative Research Centre for Sensor Signal and Information Processing, where he worked mainly in the area of adaptive algorithms for GPS interference suppression. Since 2005 he has been with the University of Adelaide, where he has been invloved in a number of research projects in the area of phased array processing for GPS, Radar and Accoustics.
|Thursday, 19th March, 2015
LG28 Napier Building (Lower Ground)