Audio Engineering Society

Chicago Section

Meeting Review, February 26, 2004

2/26/04 Meeting Highlights
by Tom Miller

Robert Zurek gave a presentation on Immersive Audio/Video Systems.� The presentation described a system that would capture and transmit 360 degree encircling audio and video streams of information.� The received signal can be manually or automatically steered to display the portion that is of greatest interest.� The Motorola PCS Research and Advanced Technology Labs have worked on concept devices applying these ideas to the areas of videoconferencing and security.�

Mr. Zurek started by tracing the history of video conferencing.� The first practical systems using satellite transmission were developed in the 1950�s, and were first publicly available in the 1970�s.� In 2000 MPEG streaming of video was possible on a wireless phone.�

Another technology that is important to immersive systems was the development of the IPIX Omniview camera in 1991.� This camera has a stationary fish eye lens that observes a 360 degree by 185 degree hemisphere.� The full view is transmitted, and a subset of the captured image is displayed at the receiving end.� The displayed image can be panned and zoomed by directing a virtual camera.� The displayed image is flattened in software, providing an image with little distortion near the center of the picture.

�

Zurek developed a similar system for the audio signal.� A circular or hemispherical array of microphones captures and transmits the complete sound field.� The observer can pan and zoom the audio signal similarly to how the picture is controlled.�

To simplify the system development, Zurek used an array of miniature cardiod microphones, and limited the bandwidth to that of traditional telephone systems.� This allowed him to use a fairly large spacing between the microphones, and to fit the camera into the center of the microphone array.�

The steering of the array was done primarily by using the microphone closest to the direction of interest and the opposing microphone.� By adding or subtracting signals from the opposing pair of microphones, he was able to obtain the full range of first order directional patterns, ranging from omnidirectional to super-caridioid.� For finer steering control, Zurek would use a weighted sum of the microphone pairs that straddled the direction of interest.�

Zurek described automatic beam steering systems he developed.� Typically the computer will be used to steer the beam towards the loudest sound.� By using non-linear processing, the beam steering signal can more precisely detect the location of the sound source.�

Zurek described how combined mode processing can be more effective.� A microphone array may first detect the general direction of a sound source, then feature identification in the video signal can be used to track the head and shoulders of the speaker person, even through short silent periods.�

He then demonstrated the operation of the system, showing the direction of the beam steering on a computer display while playing the steered array through headphones.� The demonstration system used a circular array of 8 microphones, and did not have a camera.

A question and answer period covered technical details of the developed system, and also diverted into a discussion of the seemingly irrational demand for more complex features and gimmicks on wireless phones.