Back to search

University of Patras

Analysis and optimization of audio-acoustic systems with emphasis on sound coloration and source position detection

Abstract

dc:description

This dissertation presents advanced methodologies for the visualization and processing of audio signals. The primary motivation and objective of the work are to develop approaches for the evaluation, mapping, and quantification of sound reproduction quality in rooms, as well as to enhance reproduction quality in headphones. Chapter 1 presents in detail the motivation and objectives of this dissertation and provides a brief description of the state-of-the-art technology related to the addressed topics. At the end of this chapter, the publications in scientific journals and the conference presentations made during the course of this dissertation are listed. Chapter 2 presents the theoretical background and the fundamental principles on which this work is based. Novel evaluation metrics are defined and introduced for assessing the quality of the auditory image produced by a stereophonic loudspeaker configuration in a room. A visualization methodology is proposed for generating maps of reliable auditory image reproduction. This methodology is designed to be general and applicable to any evaluation metric, not solely to the one defined and employed in this study. These methods are described in detail in Chapter 3. The evaluation framework is investigated through three distinct approaches. Central to all three is the estimation and detection of the auditory image’s direction of arrival (DoA) using binaural cues. The first approach employs a classical model of the human auditory mechanism to calculate and estimate the angle of arrival. The second approach introduces an original machine learning algorithm based on a random tree classifier, trained under ideal listening conditions. Once trained, the classifier is capable of accurately localizing the auditory image in reverberant environments and with non-ideal sound sources, achieving a high level of performance. The third approach utilizes a neural network trained on a diverse set of excitation signals and acoustic conditions. Training efficiency is enhanced through a novel input data compression technique. The method yields results comparable to, or surpassing, those of other implementations proposed in the research literature. A comprehensive description of these methods and their results is provided in Chapter 4. Chapter 5 addresses the correction of spectral distortions in music-listening headphones, with a focus on devices designed for use with portable audio playback systems. The aim is to design filters that are compact, efficient, and directly implementable on low-power digital signal processors (DSPs). For this purpose, a state-of-the-art technique for filter coefficient design is employed, enabling the creation of filters targeted toward a specified reference frequency response. This technique is integrated into a novel automatic filter design algorithm, offering control over filter order, reference response specification, and the degree of approximation to the target response. In addition to the objective performance improvements, the proposed methods have been validated through subjective listening tests. Chapter 6 details the methodology and results of these evaluations. Finally, Chapter 7 provides a concise summary of the findings and conclusions and outlines potential directions for future research and development

Degree

thesis:*
Grantor dc:publisher
University of Patras
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Kamaris, Gavriil
  • Καμάρης, Γαβριήλ

Subjects

dc:subject × 15

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/59793

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Kamaris, Gavriil; Καμάρης, Γαβριήλ. Analysis and optimization of audio-acoustic systems with emphasis on sound coloration and source position detection. University of Patras, 2025. http://hdl.handle.net/10442/hedi/59793