Perceptual Assessment of Sound Field Synthesis

H Wierstorf, Perceptual Assessment of Sound Field Synthesis, (2014). [ link ] [ paper ]

Bibtex

@phdthesis{Wierstorf2014e,
    title   = {Perceptual Assessment of Sound Field Synthesis},
    author  = {Wierstorf, Hagen},
    school  = {Technische Univsersit\"{a}t Berlin},
    address = {Berlin, Germany},
    year    = {2014},
    doi     = {10.14279/depositonce-4310},
    url     = [https://doi.org/10.14279/depositonce-4310}
}

Abstract

This thesis investigates the two sound field synthesis methods Wave Field Synthesis and near-field compensated higher order Ambisonics. It summarizes their theory and provides a software toolkit for corresponding numerical simulations. Possible deviations of the synthesized sound field for real loudspeaker arrays and their perceptual relevance are discussed. This is done firstly based on theoretical considerations, and then addressed in several psychoacoustic experiments. These experiments investigate the spatial and timbral fidelity and spectro-temporal-artifacts in a systematic way for a large number of different loudspeaker setups. The experiments are conducted with the help of dynamic binaural synthesis in order to simulate loudspeaker setups with an inter-loudspeaker spacing of under 1 cm. The results show that spatial fidelity can already be achieved with setups having an inter-loudspeaker spacing of 20 cm, whereas timbral fidelity is only possible for setups employing a spacing below 1 cm. Spectro-temporal artifacts are relevant only for the synthesis of focused sources. At the end of the thesis, a binaural auditory model is presented that is able to predict the spatial fidelity for any given loudspeaker setup.

Supplementary material

The code and links to the data are available at github. The Errata as a Jupyter Notebook.