Handbook for Sound Engineers

Virtual Systems 1451

delayed signal, while there is no delay on the side chain
input, is known as a look-ahead limiter. By setting the
delay to about three times the attack time of the limiter,
the limiter has time to react to an audio signal before that
signal reaches the limiter. Since the limiter has an attack
time of 1 ms, we will set the delay to 3 ms. A look-ahead
limiter is able to accurately limit audio transients without
the distortions inherent in ultrafast limiters.
The mid- and high-frequency outputs of the cross-
over are processed similarly. The midfrequency output
just has the parametric EQ, delay, and limiter, while the
high-frequency output has a shelving equalizer to
compensate for the CD horn, parametric EQ, delay, and
limiter, Fig. 38-11.
As you can see, while this circuit is relatively simple,
using a virtual audio processor allowed us to optimize it
in ways not possible using either commonly available
analog components or a fixed configuration digital
processor. This schematic utilizes about 3% of the
resources in the small version of the QSC virtual
processor. By way of comparison a similar schematic
took 19% of the available DSP resources on a single
MediaMatrix board. This shows the great improvement
in DSP processing speed in the latest generation of
virtual processors.
The larger and more complex the system, the greater
the advantages of the virtual audio processor over
previous technologies. Legislative chambers, stadiums,
ballrooms, theme parks, and churches are among the
facilities utilizing virtual audio processors.
One technique commonly used in legislative sound
systems is called mix-minus. Often such systems will
have a microphone and loudspeaker for each legislator.
In order to prevent feedback, each loudspeaker receives
a mix that does not contain its associated microphone
signal. Signals from other nearby microphones are at a
reduced level in the mix. The U.S. Senate sound system

utilizes this technique. Since there are 100 senators each

of whom has his or her own microphone and loud-

speaker, plus leadership microphones and loud-

speakers, there were over 100 microphones with over

100 associated loudspeakers, which would have

required over 100 mixers each with over 100 inputs if it

had been implemented with a straightforward matrix

mixer. To reduce this complexity, the mix-minus tech-

nique was developed. It works on the concept that only

a small number of microphone inputs need to be muted

or reduced in level on any given output. A single large

mixer is used to produce a mix of all the inputs called

the sum. Each output mixer receives the sum and just

those inputs that must be muted or reduced in level. The

polarities of the direct inputs of the mixer are reversed,

so that as their level is increased, they cancel out part or

all of their audio from the sum at the output of the

mixer. If a direct input is set to unity gain, it will

perfectly subtract from the sum signal, thus eliminating

that input from the mixer output. While this technique

has been used in analog system designs, circuit stability

restricted its practical use in larger systems. Digital

systems add another potential complexity. As signals

are processed and transferred between DSP processing

chips, delays may be introduced. If the sum and direct

input signals do not arrive at the mixer at exactly the

same time, the direct signal will not properly cancel.

Small amounts of signal delay on selected inputs may

be required to assure that all the signals reach the input

of any given mixer at the same time. Some virtual signal

processors automatically provide such compensation, or

provide it as an option. It is always possible to manually

insert very small delays as required.

Today, virtually all larger sound reinforcement

systems utilize some form of virtual sound processor.

The advantages of more optimized system design, the

ability to make easy changes, and reduced cost, have

Figure 38-11. Completed crossover example schematic.