Combination of localisation cues
Necessity of location detection in a width model
It is necessary to include an estimation of the perceived location of sounds within a model of the perceived auditory source width to take into account the fact that different simultaneous frequency components may have differing lateral positions. This can be demonstrated in the example shown below. If the location of each of the frequency components is not taken into account, then the overall width of the sound would be equal to the wider of the two components. As the two components are at different lateral positions, the overall width should logically be from the left hand edge of the left component to the right hand edge of the right component. A prediction of the perceived lateral position of each of the components is therefore required to take this into account.
Detection of the perceived location
As mentioned above, the main cues to the perceived lateral position of a sound are the interaural level difference (ILD)
and the interaural level difference (ITD)
. These are detected individually as discussed previously.
The two components that contribute to the perceived location are combined within the brain using a complex process that also takes into account such factors as the plausibility of the results, based on experience [Blauert 1997
]. As the main purpose of this measurement model is to predict the perceived width of the sound rather than the perceived location, a relatively simple method of combining the ITD and ILD was employed. This involves a simple trade-off equation to convert the ILD to an equivalent ITD based on a combination of the subjective data of Damaschke and colleagues [Damaschke et al 2000
] and measurements made of HRTF data.
It is recognised that this method of combining the ITD and ILD is a gross oversimplification of the process that combines these factors within the brain. Due to this, it is likely that the prediction of the perceived effect caused by conflicting cues is likely to be erroneous. However, it has been found that the results of this method are satisfactory for the large majority of signals captured with a head and torso simulator.
Once the data are converted to a common scale in terms of an equivalent ITD, they are summed and then passed on to be combined with the width measurement data.