| Abstract | Results | Contact ]

Abstract

Simulating room acoustics using numerical methods for solving the underlying wave equation yields accurate results, but becomes intractable for higher frequencies due to expensive computations. On the contrary, GA methods are efficient but are only considered accurate well above the Schroeder frequency. By combining the two methods, the full human auditory frequency range can be modeled, but two main question remains: how should the cross-over frequency between the two methods be determined and how do the room conditions influence the result? In this work, we investigate the ability of the geometrical methods to capture the underlying physics in various room conditions, translating into a lower frequency bound determining the cross-over frequency. A listening test is planned, where the cross-over frequency is determined when the sound field becomes perceptual similar enough as the objective pressures get closer to each other.

Results

These results are preliminary.

GA - Geometrical Acoustics

Rectangular room

Clap

DG-FEM GA
r2

Lo-fi drum beat

DG-FEM GA
r1
r2

Ambient sound scape

DG-FEM GA
r1
r2

Synth bass

DG-FEM GA
r1
r2

Studio (non-parallel walls)

Lo-fi drum beat

DG-FEM GA
r1

Ambient sound scape

DG-FEM GA
r1

Synth bass

DG-FEM GA
r1

L-shaped room

Lo-fi drum beat

DG-FEM GA
r1
r4

Ambient sound scape

DG-FEM GA
r1
r4

Synth bass

DG-FEM GA
r1
r4

Contact

Nikolas Borrel-Jensen, PhD student, nibor@elektro.dtu.dk
Department of Electrical Engineering, Technical University of Denmark
Oersteds Plads, Building 352, room 004, 2800 Kgs. Lyngby

Hermes Sampedro Llopis, PhD student, nhsllo@elektro.dtu.dk
Department of Electrical Engineering, Technical University of Denmark
Oersteds Plads, Building 352, room 004, 2800 Kgs. Lyngby

Supervisor: Cheol-Ho Jeong, Associate Professor
Department of Electrical Engineering, Technical University of Denmark
Oersteds Plads, Building 352, 2800 Kgs. Lyngby