Stereophonic sound reproduction for multiple listeners

Rodríguez Lizcano, Juan de Dios (2015) Stereophonic sound reproduction for multiple listeners. University of Southampton, Doctoral Thesis.

Abstract

This thesis attends to the problem of producing a common and simultaneous stereophonic illusion to two listeners by employing a pair of loudspeakers. Since the development and popularization of the stereophonic system, the deterioration of the created spatial scene when the listener departs from the symmetric midline between the loudspeakers has been recognized. Acknowledging the importance of the problem, numerous studies have tried to enlarge the listening area from different approaches without solid success.
Two known psychophysical analytical models about localization of sounds are introduced to understand the creation of phantom images as well as their extents and limitations. These models are also employed to address the problem a listener experiences outside the stereophonic?s sweet spot. Based on this analysis, a solution is proposed. Additionally to these models, an auditory localization perceptual model is incorporated to the study (CC model). The functioning of this model lies on combining binaural cues in order to produce estimates where an extension of it (CCe model) is proposed in order to cover a wider audible spectrum. Although this model is also a psychophysical analytical one, it attempts to approach to those based in physiological findings of the human auditory localization system. The intention of this model is not destined for evaluation purposes only but also to serve as basis for a filtering stage design.
A study about the subjective spatial perceptions is designed in order to recognize the scope of the CCe auditory localization model. From this listening test is concluded that the model is able to predict in a relatively good agreement the angular locations of the phantom images for frequencies up to 1 kHz. Higher frequency estimates must be considered as yet incomplete, especially for the asymmetric condition due to the used process on the extraction of group delay interaural time differences and its relation with the concepts of conversion factors and internal noise.
Several attempts based on known linear system design such as the least squares approach,which is intended to match desired pressures, are performed towards the aim of this study. In addition, several optimization methods are used in order to minimize a cost function devised from the CCe model?s perspective. The intention is to associate a well-established filtering stage design with a physiologically-approached perceptual model. It is concluded that none of the attempted methods are able to create, with a pair of channels, appropriate gains and delays across the whole audible range capable of conveying a common and simultaneous stereo illusion. The latter is due to a geometrical-frequency dependency where the appearance of a particular adjustment, along with a hyperbolic geometric notion are worthy of further research. Aditionally to the latter, the filters? design through the CCe model seems to produce relatively appropriate estimates, although only for a specific frequency range, which are not provided by an equivalent design based on the pressure matching scheme.

This record has no associated files available for download.

Contributors

Download statistics