Using a mic from long distance with voice recognition
da...@citilink.com
I want to be able to have a mic sitting in a room and have the user be
able to talk from anywhere in the room, maybe as far away as 20 feet.
Now, I am new to voice recognition and have only played with it a
limited amount with the microphone that came with my sound card.
The actual application that will be used with this set-up has not been
programmed yet, but will be based around the IBM voice recognition
engine. This will not be used for dictation originally, so I am
guessing that the input doesn't need to be as clear. I'm guessing that
with a limited vocabulary and grammar, the VR engine will be able to
recognize words, even if they are quiet or if there is noise.
Ideally, this system would work even in a noisy environment.
From what I have read about this topic, noise cancelling mics only work
when the sound is close to them. Is this true? I mean, does the gain
have to be low for noise cancelling to work?
Is there *any* solution to what I want to do?
If there isn't a mic that will work like this with voice recognition
over long distances, I'd be willing to consider a microphone that would
clip onto someone and that they would wear all the time. It would have
to be small and non-cumbersome, however. All the mics I have seen that
wireless are worn on the head, connecting to transmitter packs. Is this
as good as it gets for wireless?
Any help or opinions on this would be appreciated.
Remember, this isn't going to be used for dictation. I'm guessing that
this will make a big difference in how clean the incoming sound needs
to be. Am I right?
David Doucette
da...@citilink.com
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
David
Stefano Bodini
mmac...@tms-co.com (Malcolm MacLeod) wrote:
>Well, you can try a wireless lavalier mic. That's the type of mic that
This is interesting !
Can you suggest also where is possible to find this stuff ?
At least also schematics !
C'ya
Steve
---------------------------------
Stefano Bodini - ISDN Consultant
sbo...@clark.net
---------------------------------
George
da...@citilink.com
I looked at those microphones. I also asked around. I had some people tell
me that those microphones will pick up too much of the noise in the room. Is
that true? Also, how far away can you realistically be from one of these
microphones and still have your voice be picked-up loud enough so that it
would work with voice recognition?
David
da...@citilink.com
In article <6i7i2l$8c5$1...@usenet11.supernews.com>,
John Murphy
Based on your distance requirements, you certainly should try a pressure zone
microphone as indicated in the prior post [Radio Shack sells a condensor mic
that I've used with success in audio recording of meetings.] This might work
in a relatively noise free environment. Mount the microphone on the ceiling
in the center of the room. A PZM is less distance sensitive than other types of
microphones in that it appears to use the entire ceiling as part of the
microphone.
[A "somewhat flawed" analogy would be that the intensity of a point light
source varies in proportion to the square of the distance, the intensity of a
infinite line light source varies in proportion to the distance, and the
intensity
of a infinite plane light source does not vary with distance.] PZM microphones
are equally good at picking up wanted and unwanted sounds. Try using a stereo
equalizer to eliminate frequency bands that, based on the relative proportion
of wanted/unwanted sounds do not contribute to the success of the voice
recognition software.
Your other option of using a wireless microphone attached to your lapel or
headset should also work and would probably work better in a noisier
environment - but using a wireless microphonee seems to be too much of a
bother to me. This is in essence a "point source" and as long as you are the
closest sound source, you should be inputting a relatively high ratio of
wanted/unwanted sound.
I don't think experimenting with "noise canceling techniques" would be that
useful unless you could afford some very expensive signal processing equipment
and could identify, predict, and directly "mic" the noise sources.
Good Luck!
John Murphy
George wrote:
> Try...
>
> http://www.crownaudio.com/pzm.htm
>
> George... ( Data Concepts )
>
> David wrote in message ...
> >I want to be able to have a mic sitting in a room and have the user be
> >able to talk from anywhere in the room, maybe as far away as 20 feet.
> >
> >
> >Ideally, this system would work even in a noisy environment.
> >
> >If there isn't a mic that will work like this with voice recognition
> >over long distances, I'd be willing to consider a microphone that would
> >clip onto someone and that they would wear all the time.
> >David Doucette
> >da...@citilink.com
Birger Petterson
I paid 750.00 from B&H Photo in NYC
Birger
Terry King
As others have stated, there is not a simple solution other than
proximity to the microphone. I have used the older Radio Shack
wireless mikes (Small condenser lavaliere mike clipped on lapel and
little black box clipped on belt) with some success for command work.
But you're 'special' and 'equipped'.
The only 'elegant' solution I've seen is complex, using an array of 5
microphones and a Digital Signal Processor to synthesize a directional
noise-cancelling 'super mike' which worked for anywhere in the room. I
have never seen a commercially-available system like this; this was in
an IBM lab.
How much information do you need to pass to the system?? How much do
you need to 'roam'? Really?
For years I have had a single circuit with pushbuttons around the house
in parallel, and had up to 25 or so different commands working. I used
Morse to get different commands (A= Alarm system Arm / Disable, V=
Verbose Mode on/off, S= Shower Mode, um, Z= A to Z report of every
variable I was keeping track of (mainly to impress fellow nerd
visitors)). Not elegant, but it works. I rarely felt I was 'too far
out of touch' from a button, and they were cheap and easy to add. I use
Voice out from the system to a series of small speakers around the
house, shop, garage, bedroom (with volume controls!).
I am working on converting this to use the regular phone system, and
you might consider that also. Better for those uninterested or
unwilling to learn some Morse Code letters. (Amazing how fast my
daughter learned S 'dit-dit-dit' for Shower Mode. Although it was
always a subject to bring up with her friends. "Dad is SO weird!" )
I understand this is all homebrew stuff that's difficult to integrate
into commercial systems, but you seem willing to experiment.
I PLAN (HAckers HA is Never Finished, that's what we LIKE about it!) to
convert to using the phones (they are already all around, and the
cordless ones are even better to keep by your side in the workshop, the
garden, or poolside). Well, actually we have a Beaver Pond: the
Topsham, Vermont equivalent of 'poolside'. Anyway, phones are cool,
DTMF (aka 'TouchTone')
receiver chips are cheap and easy to use. The challenge is to wire the
phones so you can 'hit 9 and connect to the HA system'. Another
motivation I have is to be able to ask for and listen to system status
without forcing everyone else around to listen to it on the speakers.
Finally, I have a pretty good working dial-in capability to the system
now, and I don't want to have TWO different models for people to have
to learn to use the system. So I will make the transition to all phone
from Morse and Phone pretty soon.
What I Started to say: Maybe you don't really NEED voice recognition if
you have less than 20 or so commands, and little true data-input
requirements.
Terry King
Little Castle Studio
..In The Woods in Vermont
tk...@together.net
Rob Davis
What about the HomeVoice Multiroom Kit. The ad saids it is for use w/
HomeVoice, but I would imagine you could hook it up to any voice
recognition software that is looking for a single source input.
I am trying to decide how (if) to wire for HAL2000 and I was wondering
about this system. Anyone use it or heard anything about it?
Rob
"You can't plow a field by turning it over in your mind"
Joseph S. Wisniewski
John Murphy wrote:
>
> Based on your distance requirements, you certainly should try a pressure
> zone microphone as indicated in the prior post [Radio Shack sells a
> condensor mic that I've used with success in audio recording of meetings.]
> This might work in a relatively noise free environment. Mount the
> microphone on the ceiling in the center of the room. A PZM is less
> distance sensitive than other types of microphones in that it appears to
> use the entire ceiling as part of the microphone.
Creative, but not quite right. A PZM mic does two things. First, it's
sensitive. Putting the mic element in the pressure zone close to a flat
surface doubles mic sensitivity, without doubling the noise induced by
the mic element. Second, it has less reverberations. The PZM mic
essentially merges the flat surface that's part of the PZM mic with a
larger flat surface, such as a conference room table or a coffee table,
so there is only one path (talker->mic) for sound to take, instead of
multiple paths (talker-> mic and talker->table->mic) interfearing with
eachother. This is a quality that helps speech recognition a bit.
Reverberations really screw up the recognizer front end.
> Try using a stereo equalizer to eliminate frequency bands that, based on
> the relative proportion of wanted/unwanted sounds do not contribute to the
> success of the voice recognition software.
Recognizers have been smart enough to do this on their own for years. I
assume NatSpeak and ViaVoice are state-of-the-art.
> Your other option of using a wireless microphone attached to your lapel or
> headset should also work and would probably work better in a noisier
> environment - but using a wireless microphonee seems to be too much of a
> bother to me. This is in essence a "point source" and as long as you are
> the closest sound source, you should be inputting a relatively high ratio
> of wanted/unwanted sound.
>
> I don't think experimenting with "noise canceling techniques" would be that
> useful unless you could afford some very expensive signal processing
> equipment and could identify, predict, and directly "mic" the noise
> sources.
That is useful, and the stereo and TV feed into the DSP in my home
system. I've also had some pretty good results using vertical mic
arrays, usually mounted in the corners of room, at a height that is a
compromise for sitting and standing talkers. A relatively simple mic
array (no DSP, just delay lines made from lengths of PVC tubing) works
fairly well.
--
Joseph S. Wisniewski | Views expressed are my own, and don't reflect
Ford Motor Company | those of the Ford Motor Co. or affiliates.
Project Sapphire | LeMans, Daytona, Bonneville, and Sebring are
jwis...@ford.com | just races, won by people driving Ford cars!
da...@citilink.com
In article <354A1A...@ford.com>#1/1,
jwis...@ford.com wrote:
>
> John Murphy wrote:
> >
> > Based on your distance requirements, you certainly should try a pressure
> > zone microphone as indicated in the prior post [Radio Shack sells a
> > condensor mic that I've used with success in audio recording of meetings.]
> > This might work in a relatively noise free environment. Mount the
> > microphone on the ceiling in the center of the room. A PZM is less
> > distance sensitive than other types of microphones in that it appears to
> > use the entire ceiling as part of the microphone.
>
> Creative, but not quite right. A PZM mic does two things. First, it's
> sensitive. Putting the mic element in the pressure zone close to a flat
> surface doubles mic sensitivity, without doubling the noise induced by
> the mic element. Second, it has less reverberations. The PZM mic
> essentially merges the flat surface that's part of the PZM mic with a
> larger flat surface, such as a conference room table or a coffee table,
> so there is only one path (talker->mic) for sound to take, instead of
> multiple paths (talker-> mic and talker->table->mic) interfearing with
> eachother. This is a quality that helps speech recognition a bit.
> Reverberations really screw up the recognizer front end.
>
exact thing that you are saying. He meant that you can receive sound from a
greater distance, since the surface it is on is collecting sound for it. He
didn't say it was less sensitive, he said it was less *distance* sensitive.
I'm glad to hear that PZMs will help VR software. I was afraid that they
might collect too much noise, but you are saying that they actually cancel
out some of the noise that would be introduced with multiple paths.
> > Try using a stereo equalizer to eliminate frequency bands that, based on
> > the relative proportion of wanted/unwanted sounds do not contribute to the
> > success of the voice recognition software.
>
> Recognizers have been smart enough to do this on their own for years. I
> assume NatSpeak and ViaVoice are state-of-the-art.
I'm guessing that it can't hurt to help them out. :) I haven't done this,
but a good test would be to put the mic right next to something generating
low frequency noise while doing voice recognition and then removing the noise
source and doing the same voice recognition. I bet the accuracy will go up.
I hope I'm wrong, however. :)
> > Your other option of using a wireless microphone attached to your lapel or
> > headset should also work and would probably work better in a noisier
> > environment - but using a wireless microphonee seems to be too much of a
> > bother to me. This is in essence a "point source" and as long as you are
> > the closest sound source, you should be inputting a relatively high ratio
> > of wanted/unwanted sound.
> >
> > I don't think experimenting with "noise canceling techniques" would be
that
> > useful unless you could afford some very expensive signal processing
> > equipment and could identify, predict, and directly "mic" the noise
> > sources.
>
> That is useful, and the stereo and TV feed into the DSP in my home
> system.
Exactly what kind of DSP system are you talking about? And why are you
feeding the TV and stereo into it? Could you provide more information? I'm
really curious.
> I've also had some pretty good results using vertical mic
> arrays, usually mounted in the corners of room, at a height that is a
> compromise for sitting and standing talkers. A relatively simple mic
> array (no DSP, just delay lines made from lengths of PVC tubing) works
> fairly well.
What is a delay line?
And, do I'm assuming the mics are vertical so that the voices of both sitting
and standing people have the same chance of reaching the mics? Is there
anything to gain from having the mics right next to each other as opposed to
scattered through-out the room? I mean, if I were to put one in each corner
of a room, wouldn't that provide better coverage?
The only advantage I can think of for having them close to each other, is so
that if you talk near them, you can drown out noise somewhat. If they are
spaced, and if there is a loud source of noise near one of them, it is going
to battle more for attention, since you'll be across the room talking into
some other mic.
Thanks for the ideas. I had already figured on using a mic array, but it is
nice to see that others are doing it too.
> --
> Joseph S. Wisniewski | Views expressed are my own, and don't reflect
> Ford Motor Company | those of the Ford Motor Co. or affiliates.
> Project Sapphire | LeMans, Daytona, Bonneville, and Sebring are
> jwis...@ford.com | just races, won by people driving Ford cars!
>
da...@citilink.com
In article <VA.0000011...@together.net>#1/1,
tk...@together.net wrote:
>
> As others have stated, there is not a simple solution other than
> proximity to the microphone. I have used the older Radio Shack
> wireless mikes (Small condenser lavaliere mike clipped on lapel and
> little black box clipped on belt) with some success for command work.
> But you're 'special' and 'equipped'.
they work equally as well?
> The only 'elegant' solution I've seen is complex, using an array of 5
> microphones and a Digital Signal Processor to synthesize a directional
> noise-cancelling 'super mike' which worked for anywhere in the room. I
> have never seen a commercially-available system like this; this was in
> an IBM lab.
And just how did the DSP synthesize a directional noise-cancelling super mic?
And, how was the array arranged? Were they dispersed throughout the room, or
stacked on top of one another?
This was in an IBM lab where they were developing voice recognition??
> How much information do you need to pass to the system?? How much do
> you need to 'roam'? Really?
A lot of information. I'm not at liberty to disclose details, but I can say
that a lot of command and control information will be passed to the system.
As for roaming, it would be limited to one room. The room might be odd
shaped, however, and not just a regular cube.
> For years I have had a single circuit with pushbuttons around the house
> in parallel, and had up to 25 or so different commands working. I used
> Morse to get different commands (A= Alarm system Arm / Disable, V=
> Verbose Mode on/off, S= Shower Mode, um, Z= A to Z report of every
> variable I was keeping track of (mainly to impress fellow nerd
> visitors)).
*laughs* :)
> Not elegant, but it works. I rarely felt I was 'too far
> out of touch' from a button, and they were cheap and easy to add. I use
> Voice out from the system to a series of small speakers around the
> house, shop, garage, bedroom (with volume controls!).
And the feedback would repeat the command that you just issued or what?
> I am working on converting this to use the regular phone system, and
> you might consider that also. Better for those uninterested or
> unwilling to learn some Morse Code letters. (Amazing how fast my
> daughter learned S 'dit-dit-dit' for Shower Mode. Although it was
> always a subject to bring up with her friends. "Dad is SO weird!" )
She'll probably change her tune when she moves out and doesn't have the
conveniences. :)
> I understand this is all homebrew stuff that's difficult to integrate
> into commercial systems, but you seem willing to experiment.
Yeah, you could say that. ;)
> I PLAN (HAckers HA is Never Finished, that's what we LIKE about it!) to
> convert to using the phones (they are already all around, and the
> cordless ones are even better to keep by your side in the workshop, the
> garden, or poolside). Well, actually we have a Beaver Pond: the
> Topsham, Vermont equivalent of 'poolside'. Anyway, phones are cool,
> DTMF (aka 'TouchTone')
> receiver chips are cheap and easy to use. The challenge is to wire the
> phones so you can 'hit 9 and connect to the HA system'.
Yes, that is what I was wondering about. How do you plan on doing this? Is
there any sort of device on the market that could attach inline with all the
phone lines at the main line coming into the house and could either monitor
for a 9 while at the dial-tone, or monitor for a sequence of numbers and then
route the line out to whatever device you happen to be using for home
automation? I'm sure this could probably be done with a PBX, but I'm
wondering if there is a more simple solution?? Anyone??
> Another
> motivation I have is to be able to ask for and listen to system status
> without forcing everyone else around to listen to it on the speakers.
> Finally, I have a pretty good working dial-in capability to the system
> now, and I don't want to have TWO different models for people to have
> to learn to use the system. So I will make the transition to all phone
> from Morse and Phone pretty soon.
Your daughter will appreciate it, I'm sure. She'll probably still label you
as a weirdo, though. ;)
> What I Started to say: Maybe you don't really NEED voice recognition if
> you have less than 20 or so commands, and little true data-input
> requirements.
Nope. I understand what you are saying, but I really need vr in this case.
Thank you much for the feedback. I appreciate it.
david @citilink.com
> Terry King
> Little Castle Studio
> ..In The Woods in Vermont
> tk...@together.net
>
>
Rod Vernen
da...@citilink.com wrote in message <6ig09h$2t7$1...@nnrp1.dejanews.com>...
>Is
>there any sort of device on the market that could attach inline with all
the
>phone lines at the main line coming into the house and could either monitor
>for a 9 while at the dial-tone, or monitor for a sequence of numbers and
then
>route the line out to whatever device you happen to be using for home
>automation? I'm sure this could probably be done with a PBX, but I'm
>wondering if there is a more simple solution?? Anyone??
JDS Stargate supports this feature. I saw Jeff demo it at this year's
Habitech. It seemed to work quite well. I believe HAL2000 also supports
this (though with voice, not necessarily DTMF).
cjy
It's rare that a room is ever "silent." Our brains are great
at tuning out ambient noise that we don't care about and
are usually bombarded with.
Take a decently sensitive mic and plug it into a decent
tape deck. Record for a while in a "silent" room then play
it back with the volume up.
Aside from tape noise, you may notice all sorts of nature and
house noises.
In the film world, with *very* quiet Nagra tape decks and the
like, the method it to record a minute of "silence" at the
beginning of each new scene (or new set). This is so they
can remove or filter those sounds out if need be in the
editting room.
By the way, PZM mics are hardly immune to noise. Anything
that moves air will show up at a mic. I learned to record
conferences with them by placing them on the conference table
on top of a napkin or hankerchief to remove all the thumping
from people touching the table (with becomes part of the mic)
or typing or shuffling papers. I had one very good recording
of someone taking notes on a Psion (PDA).
Voice generally occupies a narrow frequency band. You can
safely knock of everything below 200HZ and above 1kHz or so.
Do this with an EQ or pick up a basic electronics book for
high-pass/low-pass filters that might cost $10 (US) to make.
Or use a much smarter DSP to create a bandpass filter and
remove a lot of quieter harmonics (reverb, and the like);
might only take a couple weeks for someone with a programming
and electronics background.
This may help a lot to eliminate furnace rumble, AC, TV squeal
(22KHz), footsteps, outside crickets, traffic and the like,
but won't help pick out one voice from a bunch of talkers.
Doing THAT requires a proximity mic (or a computer as
sophisticated as the brain).
Note that Star Trek and the like rarely has people chatting
in the background like real life. Oh yeah, and the "computer"
listening is Majel Barrett in a sound booth.
I had a friend who figured out "clapper hacking" (I *had* to buy
one :). I'd clap on the light, he'd just clap sometime in the
middle of that and confuse the poor thing.
Same idea here, but with more sophistication.
chuc...@snew.com
[remove the obvious from the From line before replying]
Joseph S. Wisniewski
da...@citilink.com wrote:
>
> In article <354A1A...@ford.com>#1/1,
> jwis...@ford.com wrote:
> >
> > John Murphy wrote:
> > >
> > > A PZM is less
> > > distance sensitive than other types of microphones in that it appears to
> > > use the entire ceiling as part of the microphone.
> >
> > Creative, but not quite right. A PZM mic does two things. First, it's
> > sensitive. Putting the mic element in the pressure zone close to a flat
> > surface doubles mic sensitivity, without doubling the noise induced by
> > the mic element. Second, it has less reverberations.
> >
> exact thing that you are saying. He meant that you can receive sound from a
> greater distance, since the surface it is on is collecting sound for it.
This, unfortunatly, is way off. The surface that a PZM is on is not
"collecting sound" at all. All the PZM is doing is operating in the
boundary layer close to a flat plane. This doubles air pressure at the
mic capsule and therefore doubles sensitivity without increasing the
electrical or thermal noises in the microphone. For speech recognition
work, this isn't an advantage, since room noise dominates over
electrical and thermal noise, even when using something as noisy as a
3mm diameter electret.
> He didn't say it was less sensitive, he said it was less *distance* sensitive.
I didn't say it was less sensitive, either. It's more sensitive. Twice
as sensitive, if you use the same mic capsule. And it's exactly as
distance sensitive as any other omnidirectional microphone.
> I'm glad to hear that PZMs will help VR software. I was afraid that they
> might collect too much noise, but you are saying that they actually cancel
> out some of the noise that would be introduced with multiple paths.
They help in that respect, but there are other ways to help mics in an
open room situation that will help more. Even a simple thing like using
multiple mics and choosing the mic with the strongest signal (microphone
diversity) will help more than trying to run with a single PZM.
The system controls the stereo and TV, so it neads to be able to hear
someone talking when the stereo or TV is on. The DSP is homebrew from
commercial DSP chips. It performs speech detection, microphone diversity
selection, LMS adaptive filtering of the TV and stereo, echo cacellation
for speakerphone use, spectral subtraction and pitch synchronous
filtering for speech recognition.
> > I've also had some pretty good results using vertical mic
> > arrays, usually mounted in the corners of room, at a height that is a
> > compromise for sitting and standing talkers. A relatively simple mic
> > array (no DSP, just delay lines made from lengths of PVC tubing) works
> > fairly well.
> What is a delay line?
Something that delays a sound. It can be a simple piece of tubing, a
delay line chip, or a DSP. By taking multiple microphones and delaying
the sound from one or more mics before combining the sound, you can make
highly directional microphones, fairly cheaply.
> And, do I'm assuming the mics are vertical so that the voices of both sitting
> and standing people have the same chance of reaching the mics? Is there
> anything to gain from having the mics right next to each other as opposed to
> scattered through-out the room? I mean, if I were to put one in each corner
> of a room, wouldn't that provide better coverage?
Mics in 4 corners of the room gives 4 times the noise without much
signal improvment if you just mix them. You can combine them with DSP
techniques (lots of DSP) or just use some simple DSP for speech
detection to do microphone diversity.
Mics close together can become a steerable microphone array, and some
simple processing lets you steer it in different directions, looking for
the strongest speech signal.
> The only advantage I can think of for having them close to each other, is so
> that if you talk near them, you can drown out noise somewhat. If they are
> spaced, and if there is a loud source of noise near one of them, it is going
> to battle more for attention, since you'll be across the room talking into
> some other mic.
That works, provided you have a reliable way of detecting speech at each
mic, so that you won't eliminate the mic near the talker and just send
one of the "noise" mics to your recognizer.
> Thanks for the ideas. I had already figured on using a mic array, but it is
> nice to see that others are doing it too.
Good luck.
John Murphy
Hi Joseph, Thanks for the clarifications and corrections. I have some questions:
I agree an array of microphones, sophisticated signal processing, and modeling known
room noise sources would be far more effective in voice recognition. Perhaps I
incorrectly assumed that these technologies are too expensive and reserved for "cold
war applications" such as phased array radar and undersea listening applications.
Are there any relatively low cost applications of these phased array technologies
with their associated signal processing requirements?.
A ceiling mounted microphone, such as a PZM, would certainly seem to minimize
relative distances (and relative sound levels) between room floor located sound
sources. This can be an advantage if you're interested in multiple sound sources,
do not have a noisy room, and have a relatively low budget. Your comments suggest
that you've had better luck selecting the loudest sound source from a collection of
directional microphones -- doesn't this still limit you to relatively quiet rooms
and require that the desired sound source be the loudest sound source?
At what point does increasing the size of the boundary (plane) behind a PZM
microphone stop contributing to the effectiveness of the microphone? I've always
looked for the biggest plane to attach the microphone (either a wall or ceiling).
I'm sure the actual microphone sound level is not exactly twice as loud but rather
varies with the size of the boundary (plane), room characteristics causing
interference patterns from secondary sound paths, sound frequency, and microphone
size and build characteristics. . .
Thanks,
John Murphy
Joseph S. Wisniewski wrote:
> This, unfortunatly, is way off. The surface that a PZM is on is not
> "collecting sound" at all. All the PZM is doing is operating in the
> boundary layer close to a flat plane. This doubles air pressure at the
> mic capsule and therefore doubles sensitivity without increasing the
> electrical or thermal noises in the microphone. For speech recognition
> work, this isn't an advantage, since room noise dominates over
> electrical and thermal noise, even when using something as noisy as a
> 3mm diameter electret.
>
> They help in that respect, but there are other ways to help mics in an
> open room situation that will help more. Even a simple thing like using
> multiple mics and choosing the mic with the strongest signal (microphone
> diversity) will help more than trying to run with a single PZM.
>
> Good luck.
>
> -