Handbook for Sound Engineers

Computer Aided Sound System Design 1369

35.2.2.1 Simulation of the Human Head

Central to incorporating the characteristics of the human
head into the simulation results and thus preparing them
for final auralization purposes is the head-related trans-
fer function. Typically, this is a data set that consists of
two directivity balloons, one for the left ear and a sec-
ond one for the right ear. Each describes, usually by
means of complex data, how the human head and the
outer part of the ear change the incoming sound waves
as they arrive at the ear. It is critical for a satisfactory
binaural auralization that the signal for each ear is
weighted with an appropriate angle- and frequency-
dependent directivity function.

The acquisition of measurement data for the human
head is not a trivial matter. Since real human heads
cannot be measured directly, a so-called dummy head
has to be built or in-ear microphones have to be used,
see Section 35.1.5.1 Human Ears. Each ear of a dummy
or a real head is equipped with a microphone. Balloon
measurements are made similar to loudspeaker balloon
measurements, only that the locations of source and
receiver are reversed and a stereo set of data files is
obtained.^27

Recent research^42 has shown that the inclusion of the
human torso into the HRTF also has significant effect
on the quality of the binaural reproduction. Even more
so, auralization results of highest quality can be
obtained utilizing a head-tracking system and a set of
HRTF balloons, where each pair of balloons describes
the transfer function for the left and right ear for a

particular angular position of the human head relative to

the human body. This data can then be employed to

auralize impulse responses of either a measured or

simulated environment with speech and music contents.

35.2.2.2 Simulation of Microphones

The need for inclusion of microphones in acoustic simu-

lation software has several reasons. On the one hand, to

be able to compare measurements with computational

results, the frequency response and the directivity char-

acteristics of the microphone have to be taken into

account. On the other hand, the possibility to simulate

either recording or reinforcement of a talker or musician

is of practical interest too. For example, by varying the

location and orientation of the pick-up microphones the

coverage can be optimized. Finally, by including micro-

phones, it becomes possible to simulate the entire chain

of sound reinforcement, from the source over the micro-

phone to the loudspeaker and back to the microphone.

Only this enables the prediction of feedback and to esti-

mate the potential gain before feedback.

However, the acquisition and distribution of micro-

phone data must still be considered in its infancy. Avail-

able data consists largely of octave-based magnitude-

only data that assumes axial symmetry. Measurement

techniques vary significantly among microphone manu-

facturers and measuring conditions, such as the

measurement distance, are not standardized and often

not even documented. Therefore most users of simula-

tion programs do not consider implementing micro-

phone data into their models, or if so, they use generic

data based on ideal directional behavior, like cardioid or

omnidirectional patterns.

There are several more issues that inhibit the wide-

spread acquisition, acceptance, and use of microphone

data.

• First, especially the measurement distance is important
  with respect to the acquisition of the data and its appli-
  cation in the software domain. A lot of microphones
  exhibit the so-called proximity effect, that is, the prop-
  erty that their frequency response and directivity func-
  tion change depending on the shape of the incident
  wave front. This effect is most visible if the acoustic
  source is within a few meters, range of the microphone
  and thus the wave front cannot be considered as plane
  anymore.


• Secondly, we described earlier with respect to loud-
  speaker data, that it is important to preserve configu-
  rability also in the software domain. In this regard,
  switchable multipattern microphones have to be

Figure 35-34. Schematic display of shadowing/ray-tracing
calculation using virtual center points. Loudspeaker boxes
are indicated by gray rectangles, sources indicated by gray
circles, virtual centers by black crosses. Only the virtual
center points are used for visibility tests.