1624 Chapter 46
- Select the microphone position. I usually begin by
looking at the on-axis response of the loudspeaker
as measured from inside of critical distance. If
multiple loudspeakers are on, turn all of them off
but one prior to measuring. The microphone should
be placed in the far free field of the loudspeaker as
previously described. When measuring a loud-
speaker’s response, care should be taken to elimi-
nate the effects of early reflections on the measured
data, as these will generate acoustic comb filters
that can mask the true response of the loudspeaker.
In most cases the predominant offending surface
will be the floor or other boundaries near the
microphone and loudspeaker. These reflections can
be reduced or eliminated by using a ground plane
microphone placement, a tall microphone stand
(when the loudspeaker is overhead), or some strate-
gically placed absorption. I prefer the tall micro-
phone stand for measuring installed systems with
seating present since it works most anywhere,
regardless of the seating type. The idea is to inter-
cept the sound on its way to a listener position, but
before it can interact with the physical boundaries
around that position. These will always be unique
to that particular seat, so it is better to look at the
free field response, as it is the common denomi-
nator to many listener seats. - Begin with the big picture. Measure an impulse
response of the complete decay of the space. This
yields an idea of the overall properties of the
room/system and provides a good point of refer-
ence for zooming in to smaller time windows. Save
this information for documentation purposes, as
later you may wish to reopen the file for further
processing.- Reduce the size of the time window to eliminate
room reflections. Remember that you are trading
off frequency resolution when truncating the time
record, Fig. 46-24. Be certain to maintain sufficient
resolution to allow adequate low-frequency detail.
In some cases, it may be impossible to maintain a
sufficiently long window to view low frequencies
and at the same time eliminate the effects of reflec-
tions at higher frequencies, Fig. 46-25. In such
cases, the investigator may wish to use a short
window for looking at the high-frequency direct
field, but a longer window for evaluating the
woofer. Windows appropriate for each part of the
spectrum can be used. Some measurement systems
provide variable time windows, which allow low
frequencies to be viewed in great detail (long time
window) while still providing a semianechoic view
(short time window) at high frequencies. There is
evidence to support that this is how humans
process sound information, making this method
particularly interesting, Fig. 46-26. - Are other microphone positions necessary to char-
acterize this loudspeaker? The off-axis response of
some loudspeakers is very similar to the on-axis
response, reducing the need to measure at many
angles. Other loudspeakers have very erratic
responses, and a measurement at any one point
around the loudspeaker may bear little resemblance
to the response at other positions. This is a design
issue, but one that must be considered by the
measurer. - Once an accurate impulse response is measured, it
can be postprocessed to yield information on spec-
tral content, speech intelligibility, and music
clarity. There are a number of metrics that can
provide this information. These are interpretations
of the measured data and generally correlate with
subjective perception of the sound at that seat. - An often overlooked method of evaluating the
impulse response is the use of convolution to
encode it onto anechoic program material. An excel-
lent freeware convolver called GratisVolver is avail-
able from http://www.catt.se. Listening to the IR can often
reveal subtleties missed by the various metrics, as
well as provide clues as to what postprocess must be
used to observe the event of interest.
Figure 46-23. A data window is used to remove the effects
of the later arrivals.
12TimeAmplitudeData window used to
isolate early energy400