Proceedings of ACOUSTICS 2016 9-11 November 2016, Brisbane, Australia ACOUSTICS 2016 Page 1 of 9 Perceived cathedral ceiling height in a multichannel virtual acoustic rendering for Gregorian Chant Peter Hüttenmeister and William L. Martens Faculty of Architecture, Design and Planning, The University of Sydney, Sydney, Australia ABSTRACT Reverberation is considered a key factor in listener envelopment. Virtual acoustic renderings of architectural spaces presented via arrays of multiple loudspeakers distributed on the horizontal plane can provide for improved listener envelopment and immersion relative to conventional stereophonic reproduction. However, with a new 22.2- channel loudspeaker format that has been submitted by NHK for an ITU recommendation, incorporating speakers at various elevations (termed height channels), there are vast new opportunities for the separation and presentation of real world acoustic events in virtual acoustic renderings. This paper presents the results of an experiment testing the hypothesis that simulated indirect sound delivered from above the listener’s head can create a distinct impression of a modelled virtual acoustic space with a tall ceiling. The experimental stimuli included spatialised indirect sound based upon an image model of the Grace Cathedral that allowed the direction and time of arrival of simulated ceiling reflections to be manipulated. A five-part, anechoically recorded Gregorian chant performance was processed to allow for spatially segregated presentation of indirect sound components (sets of discrete early reflections and global reverberation) via ear-level and height channels. Results show that listeners could distinguish well between six virtual acoustic renderings in terms of both ‘ceiling prominence’ and ‘spatial width.’ 1. INTRODUCTION Reverberation plays a vital role in spatial impression (SI), with listener envelopment (LEV) and apparent source width (ASW) as the perceptual attributes defined by the acoustical characteristics of the space. The sound that arrives at a listener’s ears is heavily influenced by a number of spatial factors, but of particular importance in the current work is the spatiotemporal distribution of reflections. Historically, in the field of concert hall acoustics, the distribution of lateral reflections has been shown to have a strong influence on SI (Barron & Marshall 1980 and Cremer 1989). Of course, in using a multichannel loudspeaker array to reproduce simulated reflections arriving from a limited set of spatial angle, the natural continuous spatial distribution is truncated in a manner that may lose important acoustical information. However, in the current work, a number of loudspeaker channels were chosen from the great number of potential loudspeaker angles represented in the 196-channel array located in the Spatial Audio Laboratory at The University of Sydney. This selection was based upon an informal evaluation performed in advance of the experimental trials in the formal investigation to be described in this paper. Previous work into multichannel by the likes of Hiyama et al (2002) and Hamasaki et al (2001) has determined appropriate positioning of loudspeakers on the horizontal plane for sufficient LEV and localisation of sound in the horizontal plane. However, with the recent adoption of elevated loudspeakers in multichannel systems that have been designed for commercial use (Hamasaki et al. 2006 and Solvang & Svensson 2006), genuine interest has been garnered around the use of these channels for reproducing immersive content considering the additional three- dimensional spatial auditory cues that they provide. Of course, listeners in a natural acoustic environment typically is provided with adequate spatial auditory cues to allow them to perceive the spatial characteristics of the space without the limited spatial resolution of a reproduction system. A room impulse response can be decomposed into three primary elements: direct sound, early reflections and late reverberation. Each of these three elements act both separately and conjointly on SI. Taking this into consideration, it might be reasonably suspected that SI is more complex than a singularly definable spatial attribute. Griesinger (1997) discusses, particularly in the context of musical performances, the relationships that exist between the musical stimulus, performance characteristics and indirect sound that influence in varying degrees overall SI. He deconstructs SI into three individual components, namely, early spatial impression (ESI), background spatial impression (BSI) and continuous spatial impression (CSI). Taking the same approach as Griesinger, this study is predominantly focused on the perception of the listener rather than the physical acoustic characteristics of the space itself. Therefore, the importance of the signal received