Skip to main navigationSkip to main content
The University of Southampton
Engineering

Research project: Compensated Amplitude Panning

Currently Active: 
Yes

A panning method that compensates for head rotation, producing stable images in any direction.

This work has improved understanding of panning reproduction, and introduced efficient improvements that make use of headtracking. These enable stable images to be produced in all directions, using minimal loudspeaker arrays.

Panning is a reproduction method where each image is created by playing the source signal through several loudspeakers, with the associated waves all arriving in sync at the listener. The gains are chosen to steer the image in different directions. Stereo is the simplest and most familiar panning system, using 2 loudspeakers. In recent years panning systems with many loudspeakers have become common. Panning can also be applied with virtual loudspeakers, which can be used in rendering binaural signals to headphones.

Panning behaviour depends on frequency. This can be understood by considering the localisation mechanisms and cues that are used by the auditory system, which have been studied using subjective experiments. In the range up to approximately 1500 Hz localisation depends mainly on two cues the Interaural Time Difference (ITD) and Interaural Level Difference (ILD). In general, panning reproduction suffers from instability. When the listener rotates their head the apparent position of images changes. Instability increases with the apparent angular separation of the loudspeakers.

 

Figure 1. The horizontal coverage using stereo CAP and VBIP systems.

Compensated Amplitude Panning takes into account the listener's head orientation to produce images that remain fixed with head rotation. It is based on a simple low-frequency spherical head model, which makes it possible to develop efficient panning formula. It turns out that this approximate model gives very good image correction in the range up to 1500Hz. More impressively, CAP can produce fixed images in any direction, including directly behind a listener facing a stereo loudspeakers. This is possible because accurate dynamic cues are generated. Current imaging systems using frontal loudspeaker arrays, such as cross-talk cancellation systems, struggle to produce rear images, so CAP is complementary to these.


CAP has been extended to produce near images, by controlling ILD. Further extensions being developed include multichannel support, channel-based decoding, and compensation for near loudspeakers.

 

Related research groups

Acoustics Group

Key Publications

Share this research project Share this on Facebook Share this on Twitter Share this on Weibo
Privacy Settings