Skip to main navigationSkip to main content
The University of Southampton
Virtual Acoustics and Audio Engineering

Research project: Object-based Room Compensation

Currently Active: 
Yes

A new method for room compensation, by the modification of reverberation data associated with sound objects.

Introduction

Conventional room compensation methods work by equalising the channel signals feeding the loudspeakers. Simpler systems can improve overall coloration in a listening region. More complex systems can invert the response, improving the time domain response to some degree. In any case it is harder to equalise the later more diffuse part of the room response. This often results in loss of clarity and a 'muddy' sound caused by excessive bass in the late response. The early and late parts of the response are often perceived as separate sound objects, so it is important these parts are each reproduced well.

In an object-based audio reproduction system, the individual sound are mixed at the point of reproduction. This concept has been extended to include the application of reverberation to the sound objects. The reverberant information is encoded in the object stream.

In this project a system has been developed that achieves a new kind of room compensation by modifying reverberation objects. It is found possible to modify parts of each reverberant response so that the important aspects of the target response are reproduced well when the modified target is played into the room. The key aspects are the early and late energies in different frequency bands, and the late decay times.

Figure 1. Ratio of reproduced clarities to the target clarity.

A useful performance measure is the ratio of the early-late energy ratios, or clarities, of the reproduced and target responses. Overall equalisation is easily achieved with existing methods. The figure shows this measure for an example in which the target and the room are both measured room responses. The measure achieves an ideal value near 1 except where the room is more reverberant than the target, indicated by target-room clarity being greater than 1. By comparison the channel-based methods perform much worse, with lower than ideal clarity.

Related research groups

Acoustics Group
Signal Processing, Audio and Hearing Group
Share this research project Share this on Facebook Share this on Twitter Share this on Weibo
Privacy Settings