Search

Consumption of audio-visual media is increasing everyday. As a result, the usage of mobile phones, tablets, and other portable devices is growing rapidly. However, in many situations it is not welcomed if the audio is not consumed via headphones within a public space. Since headphones have their own disadvantages, a proposed solution is to deliver the desired audio to the listener using loudspeaker arrays in such a way that it causes minimal interference outside of the zone of focus. Such a solution is not always practical thus, a second solution is also proposed: to enrich the audio via headphones to provide a more immersive listening experience. This thesis describes research which explored both the concept of focused delivery using a loudspeaker array, and secondly, 3D sound generation using headphones. It proposes two systems that provide immersive sound localisation either with loudspeakers or headphones, minimizing disruption to other non-users. The first system is a real-time sound projector, referred to in this thesis as a Personal Audio Delivery system, that in real time, changes the focal point of an acoustic beam with respect to the triangulated coordinates of a recognised and tracked user moving within the enclosed volume of an anechoic chamber. The Personal Audio Delivery system updates the focal point with a refresh rate of 4 Hz whilst providing an average acoustic contrast of between 23.47 dB to 34.52 dB between the focal point and rest of the anechoic chamber, combined with a triangulation accuracy of ±5 cm in a 33m^3 room. The second described system is a real-time audio virtualiser, referred to in this thesis as a Virtual Audio Localisation system. This can adaptively virtually localise the desired audio with respect to the azimuth, elevation, and distance of the triangulated coordinates of the designated speakers and listeners with an update rate of 1 Hz, to enhance the headphone listening experience by providing a realistic listening experience. Since the Virtual Audio Localisation system is integrated with the previously defined recognition and tracking system, it has a triangulation accuracy of ±5 cm in a 33m3 room.

The main focus of the research presented within this thesis is, as the title suggests, binaural spatialization. Binaural technology and, especially, the binaural recording technique are not particu-larly recent. Nevertheless, the interest in this technology has lately become substantial due to the increase in the calculation power of personal computers, which started to allow the complete and accurate real-time simulation of three-dimensional sound-fields over headphones. The goals of this body of research have been determined in order to provide elements of novelty and of contribution to the state of the art in the field of binaural spatialization. A brief summary of these is found in the following list: • The development and implementation of a binaural spatialization technique with Distance Simulation, based on the individual simulation of the distance cues and Binaural Reverb, in turn based on the weighted mix between the signals convolved with the different HRIR and BRIR sets; • The development and implementation of a characterization process for modifying a BRIR set in order to simulate different environments with different characteristics in terms of frequency response and reverb time; • The creation of a real-time and offline binaural spatialization application, imple-menting the techniques cited in the previous points, and including a set of multichannel(and Ambisonics)-to-binaural conversion tools. • The performance of a perceptual evaluation stage to verify the effectiveness, realism, and quality of the techniques developed, and • The application and use of the developed tools within both scientific and artistic “case studies”. In the following chapters, sections, and subsections, the research performed between January 2006 and March 2010 will be described, outlining the different stages before, during, and after the development of the software platform, analysing the results of the perceptual evaluations and drawing conclusions that could, in the future, be considered the starting point for new and innovative research projects.

This work is focused on the trade-off between sound quality and loudspeaker gain that occurs when using the pressure matching beamforming method to generate stereo crosstalk cancellation with uniform linear arrays.An approach to gain management is proposed that leverages characteristics of subclasses of stereo audio signals (e.g. binaural) to partially overcome this tradeoff by improving both loudspeaker gain and sound quality together.The novelty of the proposed approach is based on an observation that loudspeaker gain management, which is normally accomplished by manipulating beamforming filters, can also be accomplished by manipulating input audio signals.It is shown that decoupling gain management from beamforming filters allows it to be signal-dependent.Also, that signal-dependent gain management gives beamforming filter designers more meaningful methods for controlling both the type and amount of sound quality that is sacrificed to manage gain.A general approach to building a signal-dependent gain management algorithm is suggested.Some specific implementations are given as examples.Finally, a discussion is presented about how signal-dependent gain management potentially establishes a future research project where perceived sound quality, as opposed to numerical sound quality, is maximized.

Virtual acoustic prototyping (auralization) is the rendering of a virtual sound field that is created from the calculated acoustic response of a modeled structure. Auralization is useful in the design and subjective evaluation of buildings, automobiles, and aircraft. The virtual acoustic prototyping method in this thesis uses finite element modeling (FEM), the equivalent source method (ESM), and head-related transfer functions (HRTFs). A tradeoff exists between the accuracy of the auralization process and the number of equivalent sources (and thus computational power) that are required. The goal of this research is to validate (numerically and subjectively) a virtual acoustic prototyping method for use in structural design; this thesis illustrates the first attempt to apply the aforementioned methods to a structure that represents a typical building or automobile. The structure's acoustics were modeled using FEM, ESM, and HRTFs. A prototype of the modeled structure was built. A 36% correlation was achieved between the model and prototype. Slight variations in boundary conditions caused significant FEM error, but the data represented a typical structure. Psychoacoustic comparison testing was performed to determine the number of equivalent sources that must be used in an auralization to accurately recreate the sound field. The number was found to be dependent on the type of noise that is played to the test subject. A clear relationship between the numerical correlation of two sounds and the percentage of subjects who could hear a difference between those two sounds was established for impulsive, broadband, and engine noises.

Binaural simulation is the recreation of a three-dimensional audio environment around a listener's head. The binaural simulation of structural acoustic data would open new opportunities in virtual prototyping and simulation. By modeling the structure as an array of vibrating monopoles and applying Head Related Transfer Functions (HRTFs) to each of the sources, a binaural simulation of this type can be created. Unfortunately, this simulation method requires an extensive amount of computer power and speed for real-time simulation, more so than is available with current technology. The objective of this research is to reduce the number of computations required in the binaural simulation of structural acoustic data. This thesis details two methods of reducing the number of real-time calculations required in this binaural analysis: singular value decomposition (SVD), and equivalent source reduction (ESR). The SVD method reduces the complexity of the HRTF computations by breaking the HRTFs into dominant singular values and vectors. The ESR method reduces the number of sources to be analyzed in real-time by replacing sources on the scale of a structural wavelength with sources on the scale of an acoustic wavelength. The ESR and SVD reduction methods can be combined to provide an estimated computation time reduction of 99.4%. In addition, preliminary tests show that there is a 97% correlation between the results of the combined reduction methods and the results found with current binaural simulation techniques.

Head-Related Transfer Functions (HRTF's) describe the directionally dependent acoustic filtering properties of the head, torso, and external ear. HRTF's are typically measured at a finite number of spatial locations at a fixed radius from the head using various acoustic system identification techniques. Since HRTF's presumably contain all of the psychoacoustic cues needed to synthesize spatial sound, these functions can be used to simulate virtual or 3-D sound for auditory displays. The purpose of this thesis is to investigate the structure of HRTF data with novel 2- and 3-dimensional visualization techniques. We show how simply looking at HRTF's in these different ways can suggest how HRTF's can be analyzed and processed in innovative ways for both engineering and musical purposes. We first provide a comparative visualization study of HRTF data in the time, frequency, space, and volume domains. We demonstrate how elevation effects produced by the pinna and diffraction effects produced by head shadowing can be found in all four domains. We then show that visualizing HRTF data sets with spatial and volumetric methods exposes some artifacts introduced into HRTF's during the measurement procedure. We develop an analytical model for this procedure, show how the model accounts for these artifacts, and illustrate how these artifacts can be found in HRTF's measured by the author and HRTF's reported elsewhere in the literature. Visualizing HRTF's with spatial techniques shows that HRTF data resembles responses produced by spatially distributed sensor arrays, or beamformers. We develop an adaptive extension to an existing beamformer model for HRTF's and use the model to investigate some dynamic and cognitive phenomena associated with directional hearing. Finally, we show how interpolation methods used to create spatial domain visualizations can be exploited to compute interpolated HRTF's corresponding to arbitrary spatial locations. We describe a listening experiment designed to evaluate the perceptual quality of the interpolation algorithm. To demonstrate how interpolated HRTF's can be put to creative use, we discuss compositional techniques developed for Fishbowl, an original, binaural composition of electro-acoustic music containing moving sound sources synthesized from interpolated HRTF's. Sound examples can be found on the accompanying CD.* *This dissertation includes a CD that is compound (contains both a paper copy and a CD as part of the dissertation). The CD requires the following application: Adobe Acrobat.