Handbook for Sound Engineers

232 Chapter 9

sentation of the sound field at any specified receiver
location by using the results of the acoustical modeling
phase. Various implementations of auralization have
been put into place24,25,26,27 at the time of this writing
but because of the tremendous developments that are
taking place in the field of auralization this section will
only explore the general concepts associated with the
topic of auralization since it is safe to say that specific
imple- mentations of auralization techniques will be
subject to changes and additions dictated by new tech-
nological advances and/or by market demands.

9.3.1 The Basic Auralization Process

The basic auralization process associated with an acous-
tical model is illustrated in Fig. 9-22.

The process starts with the reflectograms repre-
senting the impulse response (IR) of the model obtained
at a specific receiver location for various frequencies.
The reflectograms contain the information pertaining to
the intensity and the direction of arrival of the reflec-
tions over a period of time that is deemed suitable to
record the desired order and length of the reflections,
and they are obtained from any of the methodologies
presented in the modeling portion of this chapter. The
reflectograms are then convolved—or mixed—with a
dry (anechoic) recording of speech or music that can be
played back under controlled conditions, using either
headphones or loudspeakers, for the purpose of subjec-
tive evaluation.

9.3.2 Implementation

The energy reaching the listener is comprised of the
direct sound, of the early reflections, and of the late
reflections as shown in Fig. 9-23.

The direct sound is easily found and modeled accu-
rately in the reflectogram since it represents the energy
traveling from source to receiver in a direct line of sight.
The only concern for the accurate auralization of the
direct sound is to insure that the attenuation follows the
inverse-distance spreading law as dictated by the source
configuration and directivity. The early reflections are
also obtained from the modeling phase but the reflecto-
gram must be limited in length—or in the order of the
reflections—because of computational constraints. The
late portion of the reflectogram is usually modeled from
a dense and random pattern of reflections with a smooth
decay and a frequency content patterned after the rever-
beration time of the room estimated at various
frequencies.

Since the reflectogram typically represents the

impulse response at a single point (or within a small

volume) in the modeled space, it must be modified in

order to represent the binaural sound that would be

reaching the eardrums of a listener by and at this point,

two separate approaches are available.

9.3.2.1 Binaural Reproduction Using Loudspeakers

The impulse response is divided into its left and right

components corresponding to the vertical left and right

planes crossing the receiver location, and thus yielding

the binaural impulse response (BIR) of the room for a

listener at the receiver location. The anechoic signal is

convolved separately for the left and the right channel,

and the result is presented under anechoic and/or near

field conditions to a listener using loudspeakers as

shown in Fig. 9-24.

This technique has the advantage of being efficient

from a computational standpoint since the process is

limited to the separation of the IR into the BIR and the

resulting convolution into left and right channels for the

playback system. The drawback of the technique is that

the playback requires a controlled environment where

the listener has to maintain a fixed position with respect

to the playback system and the crosstalk between the

loudspeakers must be very small in order to yield the

proper sense of spatial impression.

9.3.2.2 Binaural Reproduction Using Headphones

In this approach, the BIR is further modified by the

application of head-related transfer functions (HRTF)

that represent the effects that the head, torso, shoulders,

and ears will have on the sound that reaches the ear-

drums of the listener. It has been shown28,29 that these

parameters have a drastic influence on the localization

of the sound and on its overall subjective assessment.

As shown in Fig. 9-25, the reproduction system must

now use headphones since the effects of the body and

head shape of the listener have already been taken into

account. The advantage of this approach is that the play-

back system is very simple; good quality headphones

are readily available and no special setup is required.

The drawback is that the implementation of the modi-

fied BIR takes time due to the computational require-

ments for the application of the HRTF. It must also be

noted that the HRTF may not accurately describe the

specific parameters that a given listener experiences,

although current HRTF research has yielded accurate