What type of microphone should I use?

[[email protected]]

Hi all,


I was looking for some advice on recording a crowd.

What I wish to do is record a crowd singing/speaking in unison, Then,
in realtime, analyse the waveform and compare it to the expected wave
pattern. In theory the quality of the recording doesn't need to be
high but good enough to pick the general peaks and troughs of the
waveform. So, its more looking at the patterns of the waveform rather
than the details.

The crowd would vary from 20-100 people who may be spread over a 100
seat area. There will also be external background sound to contend
with.

I am assuming that the best type to use would be directional so as to
limit the external sound and to attempt to isolate the crowd. I would
also guess that using multiple mic. would be required.

Could anyone recommend a reasonable mic. for the job and how many
would you suggest using in the room(not sure of an exact size but
could be approx 10 * 10 seats with walkways and stage at front).

Apologies if its such a simple question but I don't work in the
recording field so any advice would be greatly appreciated.

Regards

Tim

[Scott Dorsey]

In article >,
> wrote:
>What I wish to do is record a crowd singing/speaking in unison, Then,
>in realtime, analyse the waveform and compare it to the expected wave
>pattern. In theory the quality of the recording doesn't need to be
>high but good enough to pick the general peaks and troughs of the
>waveform. So, its more looking at the patterns of the waveform rather
>than the details.

This doesn't seem like all that useful, but okay.

>The crowd would vary from 20-100 people who may be spread over a 100
>seat area. There will also be external background sound to contend
>with.
>
>I am assuming that the best type to use would be directional so as to
>limit the external sound and to attempt to isolate the crowd. I would
>also guess that using multiple mic. would be required.

No, try a single omni mike. EV 635A is a traditional choice for this
kind of thing.

>Could anyone recommend a reasonable mic. for the job and how many
>would you suggest using in the room(not sure of an exact size but
>could be approx 10 * 10 seats with walkways and stage at front).

There's normally no reason to use more than one mike, but I suppose it
depends on what you really want the recoriding for.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Roy W. Rising[_2_]]

(Scott Dorsey) wrote:
> In article
> >,
> > wrote:
> >What I wish to do is record a crowd singing/speaking in unison, Then,
> >in realtime, analyse the waveform and compare it to the expected wave
> >pattern. In theory the quality of the recording doesn't need to be
> >high but good enough to pick the general peaks and troughs of the
> >waveform. So, its more looking at the patterns of the waveform rather
> >than the details.
>
> This doesn't seem like all that useful, but okay.
>
> >The crowd would vary from 20-100 people who may be spread over a 100
> >seat area. There will also be external background sound to contend
> >with.
> >
> >I am assuming that the best type to use would be directional so as to
> >limit the external sound and to attempt to isolate the crowd. I would
> >also guess that using multiple mic. would be required.
>
> No, try a single omni mike. EV 635A is a traditional choice for this
> kind of thing.
>
> >Could anyone recommend a reasonable mic. for the job and how many
> >would you suggest using in the room(not sure of an exact size but
> >could be approx 10 * 10 seats with walkways and stage at front).
>
> There's normally no reason to use more than one mike, but I suppose it
> depends on what you really want the recoriding for.
> --scott

I generally agree with Scott. If the acoustics are too "live", consider an
EV RE20 cardioid. I've mic'd hundreds of audiences for their response to
TV shows. I found that placing the mic just forward of the front row and
as far above the crowd as it is wide/deep gave a good pickup. If that
height was not possible due to the size of the group, more mics were used.
For your purpose, however, the most usable waveform will come from a single
mic.

--
~
~ Roy
"If you notice the sound, it's wrong!"

[Ty Ford]

On Mon, 10 Dec 2007 11:03:08 -0500, wrote
(in article
>):

> Then, in realtime, analyse the waveform and compare it to the expected wave
> pattern.

What do you expect the expected wave pattern to be and where would you find
it so you could compare it with your recording?

Regards,

Ty Ford


--Audio Equipment Reviews Audio Production Services
Acting and Voiceover Demos http://www.tyford.com
Guitar player?:http://www.youtube.com/watch?v=4RZJ9MptZmU

[[email protected]]

Thanks for the responses guys.
>
> What do you expect the expected wave pattern to be and where would you find
> it so you could compare it with your recording?
>

I think the only way we could do this with any form of accuracy would
be to bring in enough people to run through a series of recordings to
provide a template sound.

For those familiar with the UK panto. , the "He's behind you"
idea.... and for those not familiar, I advise strongly against ever
going to watch them ;-)


>I found that placing the mic just forward of the front row and
as far above the crowd as it is wide/deep gave a good pickup.

Surely, by using an omni, it would also pick up the sound from the
surrounding speakers that are providing background ambient sound? and
thus create enough variation on the waveform so as not to be
recognised.

Would it not be better using a mic. with a more directioanl pattern
such as a cardiod suspended above to eliminate that? or is the pickup
pattern likely to be too narrow?


Thanks for your help.
Tim

[Scott Dorsey]

> wrote:
>Thanks for the responses guys.
>I think the only way we could do this with any form of accuracy would
>be to bring in enough people to run through a series of recordings to
>provide a template sound.

A template for WHAT?

>For those familiar with the UK panto. , the "He's behind you"
>idea.... and for those not familiar, I advise strongly against ever
>going to watch them ;-)

I'm really not sure what you're talking about.

>>I found that placing the mic just forward of the front row and
>as far above the crowd as it is wide/deep gave a good pickup.
>
>Surely, by using an omni, it would also pick up the sound from the
>surrounding speakers that are providing background ambient sound? and
>thus create enough variation on the waveform so as not to be
>recognised.

Wait. Wait. There are surrounding speakers now too? It's not just
the crowd, now you have changed all the rules and it's a crowd in a
room with a PA system.

>Would it not be better using a mic. with a more directioanl pattern
>such as a cardiod suspended above to eliminate that? or is the pickup
>pattern likely to be too narrow?

What the hell are you trying to do?
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[[email protected]]

Ok , might not have made myself clear.

What I need to do is record an audience shouting out in unison a
number of phrases such as : "He's behind you", "Open Sesame" ,"Speak
friend and enter" etc

Once we have a number of these phrases recorded, we will use these
waveforms as a guide pattern.
So there may be 8 phrases used with distinctive patterns.

As part of a stage show, the audience would interact by shouting out
at appropriate moments.
The audience would be quiet until the required moment (a trigger by
what's happening on stage). When they shout out, we would take that
soundwave and in realtime, compare it to the waveform guide pattern.
If its similar enough (tolerances allowed) , then an event would be
triggered, be it lights, video etc.

As there is a show occuring, external sound will be generated within
the room.We wish to compare the wave pattern of the audience shout
against the pre-recorded comparison waveform and so need to isolate
(to a degree) the audience from any background.

Effectively it is voice recognition for an audience. In theory, the
waveform should generate a similar peaks and troughs due to the
internation of the phrase and when compared would be within acceptable
tolerances for the particular phrase.

We therefore need to be able to record an audience shouting but limit
the amount of sound picked up from the surrounding so that our
waveform is more accurate for comparison purposes. This is why I was
thinking of more of a directional mic. pointed at the audience.

Hope that makes more sense.


On 11 Dec, 14:17, (Scott Dorsey) wrote:
> > wrote:
> >Thanks for the responses guys.
> >I think the only way we could do this with any form of accuracy would
> >be to bring in enough people to run through a series of recordings to
> >provide a template sound.
>
> A template for WHAT?
>
> >For those familiar with the UK panto. , the "He's behind you"
> >idea.... and for those not familiar, I advise strongly against ever
> >going to watch them ;-)
>
> I'm really not sure what you're talking about.
>
> >>I found that placing the mic just forward of the front row and
> >as far above the crowd as it is wide/deep gave a good pickup.
>
> >Surely, by using an omni, it would also pick up the sound from the
> >surrounding speakers that are providing background ambient sound? and
> >thus create enough variation on the waveform so as not to be
> >recognised.
>
> Wait. Wait. There are surrounding speakers now too? It's not just
> the crowd, now you have changed all the rules and it's a crowd in a
> room with a PA system.
>
> >Would it not be better using a mic. with a more directioanl pattern
> >such as a cardiod suspended above to eliminate that? or is the pickup
> >pattern likely to be too narrow?
>
> What the hell are you trying to do?
> --scott
> --
> "C'est un Nagra. C'est suisse, et tres, tres precis."

[Paul Stamler]

> wrote in message
...

> As part of a stage show, the audience would interact by shouting out
> at appropriate moments.
> The audience would be quiet until the required moment (a trigger by
> what's happening on stage). When they shout out, we would take that
> soundwave and in realtime, compare it to the waveform guide pattern.
> If its similar enough (tolerances allowed) , then an event would be
> triggered, be it lights, video etc.
>
> As there is a show occuring, external sound will be generated within
> the room.We wish to compare the wave pattern of the audience shout
> against the pre-recorded comparison waveform and so need to isolate
> (to a degree) the audience from any background.
>
> Effectively it is voice recognition for an audience. In theory, the
> waveform should generate a similar peaks and troughs due to the
> internation of the phrase and when compared would be within acceptable
> tolerances for the particular phrase.

Won't happen. Too many variables for a computer to trigger reliably.

On the other hand, this is exactly the sort of pattern recognition at which
human beings excel. Get a stagehand with a button to trigger the lights,
video, etc..

Peace,
Paul

[Scott Dorsey]

> wrote:
>What I need to do is record an audience shouting out in unison a
>number of phrases such as : "He's behind you", "Open Sesame" ,"Speak
>friend and enter" etc
>
>Once we have a number of these phrases recorded, we will use these
>waveforms as a guide pattern.
>So there may be 8 phrases used with distinctive patterns.
>
>As part of a stage show, the audience would interact by shouting out
>at appropriate moments.
>The audience would be quiet until the required moment (a trigger by
>what's happening on stage). When they shout out, we would take that
>soundwave and in realtime, compare it to the waveform guide pattern.
>If its similar enough (tolerances allowed) , then an event would be
>triggered, be it lights, video etc.

This doesn't really work. Voice recognition, even for vague things
like this, turns out to be very hard. Human beings are amazingly good
at it, but machines are very bad at it.

>Effectively it is voice recognition for an audience. In theory, the
>waveform should generate a similar peaks and troughs due to the
>internation of the phrase and when compared would be within acceptable
>tolerances for the particular phrase.

It won't.

>We therefore need to be able to record an audience shouting but limit
>the amount of sound picked up from the surrounding so that our
>waveform is more accurate for comparison purposes. This is why I was
>thinking of more of a directional mic. pointed at the audience.

You won't be able to do this at a performance, you're going to have to
do it in isolation. But it won't do you much good anyway, because
the pattern matching is non-trivial.

Better to have a human being juggling cues. Humans are very good at
pattern recognition. Much better than machines so far.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce]

On Tue, 11 Dec 2007 06:53:53 -0800 (PST),
wrote:

>Ok , might not have made myself clear.
>
>What I need to do is record an audience shouting out in unison a
>number of phrases such as : "He's behind you", "Open Sesame" ,"Speak
>friend and enter" etc
>
>Once we have a number of these phrases recorded, we will use these
>waveforms as a guide pattern.
>So there may be 8 phrases used with distinctive patterns.
>
>As part of a stage show, the audience would interact by shouting out
>at appropriate moments.
>The audience would be quiet until the required moment (a trigger by
>what's happening on stage). When they shout out, we would take that
>soundwave and in realtime, compare it to the waveform guide pattern.
>If its similar enough (tolerances allowed) , then an event would be
>triggered, be it lights, video etc.
>
>As there is a show occuring, external sound will be generated within
>the room.We wish to compare the wave pattern of the audience shout
>against the pre-recorded comparison waveform and so need to isolate
>(to a degree) the audience from any background.
>
>Effectively it is voice recognition for an audience. In theory, the
>waveform should generate a similar peaks and troughs due to the
>internation of the phrase and when compared would be within acceptable
>tolerances for the particular phrase.
>
>We therefore need to be able to record an audience shouting but limit
>the amount of sound picked up from the surrounding so that our
>waveform is more accurate for comparison purposes. This is why I was
>thinking of more of a directional mic. pointed at the audience.
>
>Hope that makes more sense.

If you have one person enunciating really carefully, this may work
perhaps 75% of the time. With a baying mob you have absolutely no
chance. Buy someone a case of beer and have him sit in the wings with
a few buttons.

d

--
Pearce Consulting
http://www.pearce.uk.com

--
Posted via a free Usenet account from http://www.teranews.com

[Richard Crowley]

> wrote ...
> Ok , might not have made myself clear.
>
> What I need to do is record an audience shouting out in unison a
> number of phrases such as : "He's behind you", "Open Sesame" ,"Speak
> friend and enter" etc
>
> Once we have a number of these phrases recorded, we will use these
> waveforms as a guide pattern.
> So there may be 8 phrases used with distinctive patterns.
>
> As part of a stage show, the audience would interact by shouting out
> at appropriate moments.
> The audience would be quiet until the required moment (a trigger by
> what's happening on stage). When they shout out, we would take that
> soundwave and in realtime, compare it to the waveform guide pattern.
> If its similar enough (tolerances allowed) , then an event would be
> triggered, be it lights, video etc.

Voice-recognition under ideal circumstances with a
single articulated talker is not terrlbly reliable. Trying
to do voice recognition of a big crowd of random
people in a live,noisy ambient is just insane. Forget it.

[Richard Crowley]

"Richard Crowley" wrote ...
> > wrote ...
>> Ok , might not have made myself clear.
>>
>> What I need to do is record an audience shouting out in unison a
>> number of phrases such as : "He's behind you", "Open Sesame" ,"Speak
>> friend and enter" etc
>>
>> Once we have a number of these phrases recorded, we will use these
>> waveforms as a guide pattern.
>> So there may be 8 phrases used with distinctive patterns.
>>
>> As part of a stage show, the audience would interact by shouting out
>> at appropriate moments.
>> The audience would be quiet until the required moment (a trigger by
>> what's happening on stage). When they shout out, we would take that
>> soundwave and in realtime, compare it to the waveform guide pattern.
>> If its similar enough (tolerances allowed) , then an event would be
>> triggered, be it lights, video etc.
>
> Voice-recognition under ideal circumstances with a
> single articulated talker is not terrlbly reliable. Trying
> to do voice recognition of a big crowd of random
> people in a live,noisy ambient is just insane. Forget it.

Your only chance is to use pre-set phrases with different
number of syllables. Then count the peaks. That technique
was used by toy-makers in the past, with varying success.
But with a big audience of random people, the sound may
not be coherent enough to even do that.

[[email protected]]

On Dec 11, 10:20 am, "Paul Stamler" > wrote:
> > wrote in message
>
> ...
>
>
>
> > As part of a stage show, the audience would interact by shouting out
> > at appropriate moments.
> > The audience would be quiet until the required moment (a trigger by
> > what's happening on stage). When they shout out, we would take that
> > soundwave and in realtime, compare it to the waveform guide pattern.
> > If its similar enough (tolerances allowed) , then an event would be
> > triggered, be it lights, video etc.
>
> > As there is a show occuring, external sound will be generated within
> > the room.We wish to compare the wave pattern of the audience shout
> > against the pre-recorded comparison waveform and so need to isolate
> > (to a degree) the audience from any background.
>
> > Effectively it is voice recognition for an audience. In theory, the
> > waveform should generate a similar peaks and troughs due to the
> > internation of the phrase and when compared would be within acceptable
> > tolerances for the particular phrase.
>
> Won't happen. Too many variables for a computer to trigger reliably.
>
> On the other hand, this is exactly the sort of pattern recognition at which
> human beings excel. Get a stagehand with a button to trigger the lights,
> video, etc..
>
> Peace,
> Paul

Agreed. Your idea sounds like an unnecessarily complicated solution to
a simple problem (cueing stage events).