You are on page 1of 13

Uses for the Spectrograph

SMAART’s spectrogram is an excellent way of visualizing old problems in new ways.


With a little bit of creativity a problem can be viewed from a different perspective.
Sometimes this is just what we need to think outside the box.

Feedback:
Feedback is complex phenomena, but its effects are well-known and not at all complex.
It basically ruins any sound performance from an airport page to a full-blown stadium
concert. A typical scenario is when an artist passes by a loudspeaker and their
microphone feedbacks. The FOH or monitor operator uses one or more tools to fix the
problem: the operator’s ears, a graphic equalizer, and a RTA. Previously, people used
their knowledge and experience to grab and cut the appropriate frequency slider on their
graphic EQ. Soon, technology evolved to produce cost-effective RTAs that were more
efficient tools that allow visual determination of problem frequencies. Operators came to
rely on it because of its repeatability and because it was a faster method. RTAs soon
included a peak hold so that a marker at the feedback frequency would be left on the
display for a couple of seconds. This gave the operator both more time to determine the
problem frequency and more reliability in grabbing the correct frequency slider. Because
the 1/3 octave graphic equalizer with frequencies on ISO centers became the de facto
standard, RTAs were designed to analyze the energy in corresponding 1/3 octave
frequency bands.

The limitation of the RTA is in finding the feedback energy by locating the small marker
on the display. The obvious problem is there is not one feedback marker but 31 markers
(See Figure 1) and the feedback problem has to be fixed while still continuing to mix the
performance. A better tool for this would be the spectrograph.
From a distance the visual representation of the feedback is more apparent using the
spectrogram than with the RTA. The display also allows feedback to be shown on the
display for a future reference. However similar to the RTA once the data has gone off the
screen it’s gone for good. Using the

A not so obvious measurement would be to use the spectrograph to measure “power


alley”. Power alley typically is used to describe the bass build-up between separated
subwoofers where their outputs sum together. A typical set-up is with subwoofers on the
left and right of a stage. Interactions will be obvious on the spectrogram, especially
where their outputs are nearly equal in level. Using the spectrograph function a
microphone was moved along the horizontal plane from one side to the other. The
beginning of each resulting spectrogram is one side of the audience areas and the end is
the other side. This will clearly show any interactions between subwoofers. While the
spectrogram cannot tell you how to fix any problems, it is the best tool to use for
demonstrating the results of before and after optimization.

These examples use small omni directional loudspeakers (See Figure 2) to emulate large
subwoofers. The emulation is achieved through simple frequency/wavelength scaling.
Thus, what happens with these devices measured at higher frequencies can be scaled to
any frequency range by maintaining the same physical separation distance to sound
wavelength ratios.
By walking the microphone through the listening area of interest, real time polars were
made using the setup shown in Figure 1. While this was in a controlled, laboratory
condition, the same thing can be done in the field by moving a microphone from one side
of the audience or stage area to the other. The measurements in Figure 2 clearly show the
shows the effects of a polarity inversion on one loudspeaker. In the left spectrogram,
there is a power alley on the centerline between the loudspeakers. On the right there are
power alleys on either side of center with a cancellation on the centerline. This is the
expected results and is as audibly significant as it is visually significant.

The effect reducing the power allies by amplitude shading is shown in Figure 4. The
same test was performed. While the first pass of the microphone through the listening
area shows the expected interaction, the second pass shows a clear reduction by using a -
6dB attenuation on one loudspeaker. While the overall level where the loudspeakers sum
is reduced by 2.5 dB, the severe cancellations across the listening area between those
points are almost eliminated. For subwoofers the loss of a less a 2.5 dB in level in
specific areas in order to provide a much more uniform across the entire area would be an
excellent compromise.
The effect reducing the power allies by frequency shading is shown in Figure 5. The left
spectrogram is with both loudspeakers in polarity, the middle shows results using
amplitude tapering as in Figure 4. The right spectrogram shows similar results using
frequency shading achieved by simple EQ adjustments. In this case the adjustement was a
½ octave wide filter with 6 dB of attenuation. In this case, the frequency was about 1.5
kHz, but similar result could be achieved in other frequency ranges with the loudspeaker
spacing scaled appropriately..
The effect of on power allies using delay tapering is shown in Figure 6. In the leftmost
spectrogram the loudspeakers no signal delay is used. In the center spectrogram, signal
delay is used on one loudspeaker. In the rightmost spectrogram the delay is increased.
Again, the power and cancellation allies shift and change. While the results are different,
they may not look that much better using this technique alone. However, what this clearly
points out is that delay certainly has a significant effect on the pattern and thus a
potentially useful role in correcting interaction problems.

Arraying the Un-Arayable:

SMAART Live 5s spectrograph can show the interaction between arrays or clusters of
loudspeakers. Because the spectrograph has no noise immunity or direct to reverberant
information, its primary usefulness is analyzing before and after results. In this a 2-wide
cluster (See Figure 7) with the loudspeakers arranged in parallel is investigated using the
various techniques of level, frequency and delay shading already discussed.
The two spectrograms in Figure 8 show the same conditions as in Figure 2. The polarity
is inverted on one loudspeaker for one of these spectrograms (the left one, in case you
could not tell which one).
In Figure 9 the left most in this series of spectrograms has no optimization. The second
has amplitude tapering. The third has delay tapering. The fourth has frequency shading.
And last we se the results of all three combined. While the combination certainly did not
eliminate problem areas, the overall results show a far wider listening area for all
frequencies and the cancellation allies are a bit narrower and somewhat fragmented (a
good thing) unlike the continuous and wider cancellation allies in the leftmost
spectrogram.

In this case, application of all three types of shading results in a more random and overall
fuller pattern. However, these measurements were essentially anechoic. With full size
loudspeakers in real rooms, these techniques can provide even more dramatic results.

Here are some examples of spectrograms from real rooms.


Here we have a typical shoe box room. The sound system consists of a stereo line array
with point source loudspeakers covering the outer sides. The green arrow indicates the
direction that the microphone was moved. This is the horizontal plane.
Another shoe box room however this time the spectrogram was made in the vertical.
This is a very uneven response. The system adjustments for the line array included
frequency shading, amplitude tapering, and delay tapering. This caused the array to loose
its cylindrical wave front. Below is the solution.

No matter how large the room a spectrogram is worth a thousand words. This is a 3500
seat hall with line arrays, and delay zone loudspeakers with a horizontal spectrogram.
Spectrograms aren’t just useful for looking at loudspeaker arrays. They can also be used
to investigate microphone arrays. Four microphones were summed into the measurement
channel of SMAART. A small loudspeaker with the Q of the human voice was moved in
the horizontal plane.

Nice even coverage.


Uneven coverage due to an inverted polarity on the 2nd microphone.

Keep in mind that the spectrogram has no time information, and doesn’t benefit from
noise immunity. However a spectrogram created pre and post optimization is very useful
just keep in mind about the reflections that exist in your measurement environment.

You might also like