Improving accuracy of sound reflection estimation using neural networks

Bachelor thesis (2023)

Authors

E. Scholtens Electrical Engineering, Mathematics and Computer Science

Contributors

E. Eisemann Computer Graphics and Visualisation - EEMCS (mentor)
Jorge Martinez (graduation committee member)
Faculty
Electrical Engineering, Mathematics and Computer Science, Electrical Engineering, Mathematics and Computer Science
Copyright: © 2023 Ekko Scholtens
More Info
expand_more

Published Date

03-02-2023

Language

English

Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Faculty

Electrical Engineering, Mathematics and Computer Science

Abstract

In this paper, we present a study to improve using neural networks for acoustic reflection localization. Our study focuses on the reimplementation of the proposed neural network model by Bologni et al. and investigates the effect of adding a third microphone to the microphone array. We reimplemented and trained the neural network using the same framework and hyperparameters as the original model, and then evaluated it using the same metrics. Our results show that the addition of a third microphone improves the amount of detected sources from 43% to 58%, it also improved the front-back ambiguity from 25% to 18%. Conclusively, have demonstrated the potential benefits of adding a third microphone to the neural network approach for acoustic reflection localization.

Files

CSE3000_Final_Paper_Ekko_schol... (pdf)
(pdf | 0.601 Mb)
Unknown license