Specific outcomes from the project

Signal properties

Division of source and environment related aspects based on time or perceptual streaming


The division of source and environment related aspects of the signal based upon a single time point does not match perception well. More accurate results can be obtained by dividing the signal based upon perceptual streaming, separating segments perceived to be either a source or reverberation and measuring these individually.


The most common IACC-based measurements of spatial impression also recommend dividing the impulse response at a fixed time interval after the arrival of the direct sound (usually 80 ms) in order to separate the source-related and environment (or reverberation) related components. Previous research by others have indicated that this time division is meaningless, and reflections arriving either side of this division can affect the spatial impression of either the source or reverberation.

A subjective experiment was conducted to verify this, and to investigate more suitable methods to separate the source and environment related aspects of the sound. This showed the following factors. Firstly, altering the characteristics of the reflections arriving with 80 ms of the direct sound caused a negligable difference to both the source and the environment width of all stimuli. Secondly, altering the characteristics of the reflections arriving more than 80 ms after the direct sound affected both the source and environment width of all the stimuli, though the latter was affected more.

As an alternative, the stimuli were also measured using perceptual grouping - where the source-related and environment-related components are separated by whether they sound to be part of the source or whether they sound to be separate reverberation. This was achieved using a relatively simple division at the point at which the direct sound ceased to arrive at the receiver. It was apparent that IACC-based measurements made using this method matched the subjective results more accurately than using the previous single time division method.


[Mason et al 2002]