Virtual Systems 1453sound systems to be designed without visible micro-
phones, even in difficult acoustic environments.
Since the outputs of the individual microphone
elements in an array are processed into the single virtual
microphone output by DSP processing, adding more
DSP will allow additional virtual microphones to be
produced from the same array. Each virtual microphone
can be aimed in a different direction, with its own direc-
tional pattern. We already see the beginnings of this in
some of the microphone systems that provide 5.1
surround outputs for recording from a single compact
microphone array. In a speech reinforcement system,
each virtual microphone could track an individual
talker. Talkers would be identified by their unique indi-
vidual voiceprint. When there are multiple microphone
arrays in the room, each talker’s individual virtual
microphone can automatically be formed using the
optimum nearby array. As each talker moves around the
room, his personal virtual microphone will always be
formed using a nearby array, and will move from array
to array as he moves around the room.
Since each virtual microphone will stay with its
assigned talker, the output of each microphone may be
individually optimized for the person to which it is
assigned. When logging of the activities in the room is
required, if desired each person could be recorded on
her own individual track. Where speech to text conver-
sion is utilized, having a separate virtual microphone for
each talker is a significant advantage. Speech to text
conversion is much easier when the system can learn
the voice of a single individual. By providing outputs to
the speech to text system that only contains the voice of
a single individual, accuracy is greatly improved.
Microphone arrays will also have the ability to selec-
tively reject sounds coming from certain sources.
During system setup, the virtual microphone processors
will be taught the location of the system loudspeakers,
and of any significant noise sources in the room. This
will allow them to keep a null in the directional pattern
always aimed in those directions. As a result, the
chances of feedback and the pickup of noise will be
significantly reduced.
It will also be possible to define regions in 3D space
from which speech will not be amplified. In legislative
systems, for example, it is extremely important to make
sure side conversations are never amplified. By defining
an area slightly back from the desks as a privacy zone,
the legislators will be able to lean back and have a
private conversation with their aides even if they forget
to turn their microphones off.
Current voice tracking microphone arrays are limited
in their bandwidth, add significant signal latency, and
are costly. These factors have made them unattractive
for sound reinforcement applications. However,
improvements in processing algorithms, coupled with
the dramatic reductions in the cost of DSP processing
power we have seen each year, will soon bring this tech-
nology to a host of new applications including sound
reinforcement.38.4.2 LoudspeakersMany loudspeakers today are powered with integrated
power amplifiers and crossovers. Some loudspeakers
have expanded on this concept by directly accepting digi-
tal audio and control signals. They contain DSP process-
ing, which, integrated with the loudspeaker system
design, allow much improved loudspeaker performance
and protection. Modern DSP-based line array loudspeak-
ers have steerable directional patterns, and in some cases
can produce multiple acoustic output beams from the
same loudspeaker. They may even send back an audio
sample of their acoustic output for confidence monitor-
ing.
As with microphone arrays, DS-based loudspeaker
arrays allow sound to be steered to where it is needed,
and kept from where it is not wanted. Dynamically
controlled loudspeaker arrays will allow the loud-
speaker coverage to change as room and system condi-
tions change. Loudspeaker arrays may be produced as
lines or flat panels, which mount flush with the walls,
ceilings, and other architectural room elements. No
longer is it necessary for loudspeakers to be aimed in
the direction we wish the sound to go. For example, it is
quite feasible to mount a flat panel loudspeaker array in
a convenient location on the sidewall of the room, and
direct the sound downwards and back into the audience
area. Loudspeaker coverage patterns and directions may
be changed under the control of the virtual system for
different uses of the facility. This is a tremendous
advantage over the older technolog, which required
either multiple sets of speakers, or physically changing
the loudspeaker aiming for different applications.
A single loudspeaker array may be used to simulta-
neously produce multiple sound coverage patterns, each
of which may be driven by its own independent sound
source if so desired. One application of this technique
would allow greatly enlarging the area in a room where
accurate multichannel reproduction could be heard.
Those located towards the edges of the room could now
receive properly balanced sound from all loudspeakers
in the room, even though they were much closer to
some of the loudspeakers than to others, thus preserving