Page 50 - Winter 2020

ROOM-ACOUSTIC AURALIZATION
ear pinnae, head, and even the torso, individual subjects can be conveniently scanned to establish finite-difference models for wave-based simulation of the head-related impulse responses (Prepelita et al., 2020). Figure 2 illustrates a 3D printed head replica using the so-created mesh model to valid meshing accuracy.
Room-Acoustic Modeling Techniques
The aim of the room-acoustic modeling is to compute RIRs in the given space as accurately and as efficiently as possible under given constraints. These constraints depend on the needs of the application, available computational resources, and so on. The range of applications is wide, starting from the acoustic design of concert halls (Hochgraf, 2019) where accuracy is crucial and computational performance is only secondary to room-acoustic research, and ending in computer games where the situation is completely opposite and real-time performance is required at the cost of low accuracy (Raghuvanshi et al., 2007).
There are many different modeling paradigms, each having their own pros and cons. But what is common to them is that they aim to produce RIRs. Figure 3 schematically depicts an energy RIR, a so-called echogram in a room. In the early part, the energy RIR encompasses direct sound and early reflections,
followed by a late reverberation part. We now discuss the main modeling principles and techniques (briefly summarized in Table 1).
Wave-Based Modeling
The wave-based modeling techniques are the most accurate models, although they are also the most inefficient because their computational loads typically depend on the frequency range to be covered. As the frequency goes higher, more computational resources are needed. As a result, they are computationally extremely expensive at the high end.
Figure 3. Echogram of an enclosed space over time from the start of the sound (left) to when it dies out (right). This is also called the energy room-impulse response. The late reverberation contains densely populated decaying reflections. Reproduced from Xiang et al., 2019, with permission.
Table 1. Capabilities and properties of two room-acoustic modeling techniques
Methods
Wave Based
Geometrical Acoustic
FDTD/FEM/BEM
Image Source
Ray Tracing
Radiosity
Transport
Reflection Type
Energy/pressure
Pressure
Pressure
Energy
Energy
Energy
Specular
Yes
Yes
Yes
No
Yes
Diffuse
No
No
Yes
Yes
Yes
Edge diffraction
Inherent
Extension
Extension
Extension
Inherent
Accuracy
Main factor
Grid density
N/A (is exact)
Number of rays
Grid density
Grid density
Computational Load
wrt Frequency
Polynomial
Constant
Constant
Constant
Constant
wrt Time
Linear
Exponential
Linear
>Linear
Linear
FDTD, finite-difference time-domain method; FEM, finite-element method; BEM, boundary-element method; wrt, with respect to. The varying capabilities and properties depend on the modeled quantities (pressure or energy), on the capability of involving scattering or other phenomena, and on the main factors affecting their accuracy and computational complexity.
50 Acoustics Today • Winter 2020