Abstract
On its path to the ear the spectral envelope of a sound is modified by transmission channels. These modifications can distort, or colour, sound timbre preventing recognition. Research by Olive et al. (1995) suggests that perceptual mechanisms remove spectral colouration caused by loudspeakers and rooms. This compensation is apparent when listening in the real-world but not in laboratory tests and may be a result of the longer time courses involved in real-world listening. Experiments conducted as part of this thesis confirm that compensation for loudspeaker/room colouration occurs using a real-world listening experimental paradigm and is perceptually moderate to large. This is partly caused by mechanisms that are sensitive to the time gaps between hearing different loudspeakers/rooms, common in real-world listening, and partly due to mechanisms that are not sensitive to these time gaps. A research process was set out to further investigate mechanisms behind the time-gap sensitive component of real-world compensation. A literature review of mechanisms that might explain this compensation was undertaken. A peripheral enhancement mechanism and a central spectral compensation mechanism cause compensation for spectral distortion caused by vocal tract (VT) characteristics in speech perception. These mechanisms are time-gap sensitive and were shown to have a number of features that mean they have the potential to cause the time-gap sensitive real-world compensation for loudspeakers and rooms. However, mechanisms that compensate for VT when listening to speech may not compensate for other channels when listening to non-speech. Laboratory tests were conducted to show that enhancement and spectral compensation also occur with non-speech sounds and therefore have the potential to contribute to any time-gap sensitive compensation for loudspeakers/rooms when listening to non-speech as well as speech. Therefore, these mechanisms can explain the real-world compensation seen in Olive et al.’s (1995) work and in real-world listening more generally. So far the specific mechanisms of real-world compensation have only been measured using laboratory studies. A framework is proposed for future work to confirm that these mechanisms explain real-world compensation using the real-world listening paradigm.