Impulse Response Recording in the Field

[fabiano83]

is there any easy way to record an impulse response on the field?
I'm asking this because i'm going to a trip school in may and we are going to visit the most important north italian theaters, such la scala in milan and la fenice in venice. since my lic. phil. dissertation is about acoustics of theaters it would be useful for me to record an IR of the theaters, so that i could use the measurements in my dissertation. The problem is, i can easy carry around a zoom H4 with a behringer ecm8000 to record, but it wouldnt be too easy to go around italy with a yamaha hs80m to play the sweeps... so what i'm asking is if there is any possibility to reproduce an IR tone with something i could easily carry around all day in my pocket or in a small bag... otherwise i won't buy the zoom since it would be pretty useless to record just the reverb of people singing or speaking. (we are going to visit the teathers in the morning/afternoon, just in case you think i was going to do sweeps during the concerts )

thanks in advance,

Fabiano

[BLP]

While you're right that a sine sweep through a nice, flat transducer would be optimal, you could make life easier by using a short-time impulse. My first thought was a starter's pistol. If they don't mind if you shoot a blank or two, you could use that excitation signal. Not the best signal-to-noise ratio and there will be some inherent frequency anomalies in each and ever blank, but it's better than nothing and quite portable. I'm assuming that, for your purposes, that could provide the data you need for your analyses later on.

[mohthom]

Pop a balloon instead of a starter pistol. People tend to be much more tolerant when you don't carry around pseudo-weaponry.

MohThoM

[BLP]

Originally Posted by mohthom

Pop a balloon instead of a starter pistol. People tend to be much more tolerant when you don't carry around pseudo-weaponry.

True, but there are a few extra complications with the balloon popping. First of all, the S/N ratio is not as good, as a balloon pop does not generate the same level of acoustic power that a starter pistol would. Also, the consistency of the balloon, the amount of air inflating it, the temperature of the air inside it and the popping implement can all have a profound effect on the acoustic energy created upon popping it.

That being said, a balloon thing will indeed work.

[mohthom]

I still find it much easier to pop a couple of extra balloons than to get a firearms license.

MohThoM

[BLP]

Originally Posted by mohthom

I still find it much easier to pop a couple of extra balloons than to get a firearms license.

MohThoM

Fair enough. I guess I'm just used to the good ol' US of A/wild wild west. Gotta love it!

[David Spearritt]

Popping a balloon will not work in LaScala. It will not work in anything other than a bathroom or a closet. Its one of those quaint bits of folklore in the industry that gets perpetuated by those that have probably not done any real measurements or had to use the results.

I recently completed some IR measurements in our local concert hall and still had trouble getting a decent S/N with a Meyer speaker and a 200W amp driving a sine sweep from Arta.

IR measurment cannot be done with anything you can carry in your pocket! Forget it. Plan a real measurement exercise, get the ARTA software and a very powerful sweep source if its a large hall. Then you have a chance.

[BLP]

Originally Posted by David Spearritt

Popping a balloon will not work in LaScala. It will not work in anything other than a bathroom or a closet. Its one of those quaint bits of folklore in the industry that gets perpetuated by those that have probably not done any real measurements or had to use the results.

I'm not sure what you were insinuating about who, so I'll chime in...

I don't think that "ideal" was a prereq by the OP. Sure the system used by the project I'm currently working on uses some 36 drivers of various shapes, sizes and enclosures, but for the purposes of measurement of basic acoustic features (T60, bass ratio, lateral energy, etc.) in a medium to small room, a balloon can work in a pinch. Since the OP is going for an extracurricular bit of data collection and, as the original post would seem to suggest, not for use as IR reverb or as the focus of a project, MohThoM and I were merely being trying to be helpful.

As for a proper S/N ratio using a less powerful excitation signal, while not ideal, there are ways around this for certain applications (i.e. not so much for convolution in audio processing, but for measurement systems). Assuming the hall noise being dealt with is stochastic (or displays a Gaussian or other predictable distribution), post-processing based on a sampling of the noise (taken before or after the excitation and decay) can help improve the S/N. The best route is always to beat it on the front end, but there are tricks...

[David Spearritt]

Hi Brett. Yes I was restricting the emphasis to IR measurement as the OP requested, although even T60 will be difficult or impossible in the 3000 seat LaScala with a balloon to get enough energy and level.

In my tests, gaining enough S/N for use in an IR plugin, one needs a large SPL for a significant period of time.

Attached is a real measured 44/24 IR synthesized from two mono sweeps in my local concert hall, 2500 seats, measured last November. The S/N is quite good, although still not quite high enough for adding a decent amount of reverb in a plugin.

LaScala is even larger than this hall.

[David Spearritt]

Fabiano

You should contact one of your countrymen, who is an expert in this field, Angelo Farina. Here is one of his early papers on the subject:
http://www.acoustics.net/objects/pdf...e_farina03.pdf

He explains in some way why the balloon does not work well. He has also done work in LaScala, so knows what you are up against.

[David Spearritt]

Fabiano, there are some nice little videos here that will give you some idea of the techniques in IR measurement.
Altiverb 6 demo page

[mohthom]

From the paper . . .

The results have shown that the differences obtained using different measurement techniques and
equipment are not of significant importance. The only effective differences were found regarding the
recording techniques, as the monoaural measurements give appreciably different results from the
average of left and right channel of binaural measurements. Slightly different results were also found
between true impulsive sources (pistol shots, balloons) and omnidirectional (dodechaedral)
loudspeakers.

In fact, the paper only ever mentions balloons in the same context as a pistol shot and doesn't differentiate between them in terms of preference - unless my skim-read hasn't been as accurate as normal . . . and the only time they're referred to separately is in the graphs at the end!

FWIW, the graph [Fig. 4] states that there's less than 3db difference in SNR between the pistol and the balloon.

I agree that it's not perfect (David, for my last IR recordings I used a Genelec 1037a with a 7070), but given the constraints it's not a bad compromise IMHO;

is any possibility to reproduce an IR tone with something i could easily carry around all day in my pocket or in a small bag

MohThoM

[David Spearritt]

Yes, some of the papers conclude that balloons could work, but are generally not specific about the following which makes me suspicious:
1. Size and volume of room for measurements (very important)
2. Type/strength of balloon/pistol, pressure or volume filled etc
3. If the response measurement was at least a critical distance from the balloon/pistol.

I have tried balloons a few times in large auditoria and in rail tunnels of all places and got completely hopeless results. The shape of the balloon pop does not give a flat spectrum and very little energy at low frequencies. Its also completely inconsistent between balloons. The signal to noise at the energetic frequencies is sometimes only 20-30dB above background and this is not what I would call enough for a solid measurement.

Its certainly not enough to create an IR file for a convolution plugin for mastering. This is why I consider them to not work in a practical sense.

[mohthom]

Originally Posted by David Spearritt

Its certainly not enough to create an IR file for a convolution plugin for mastering.

Then we're good, no?

For a 'Mastering IR' I wouldn't dream of using anything other than a full-range fairly flat loudspeaker or array in conjunction with a decent quality measurement microphone. That's not the goal for the OP.

Also, I'd really like to see the paper you refer to that details why balloons are no good - what can I say, I'm a glutton for technical reading based punishment! I'd be particularly interested in any papers you know of that discuss different sizes and inflation pressures, as this would seem to me to be of paramount importance.

MohThoM

[David Spearritt]

The reference from Angelo Farina quoted above shows the S/N graph of MLSSA vs Balloons. Audio Ease has two papers that confirm the same results, that impulse excitation is inferior.

http://www.audioease.com/Pages/Altiv...or%20music.pdf

http://www.audioease.com/Pages/Altiv...20IR's.pdf

I guess I wrongly assumed that the OP was after IR in LaScala (what I thought I read), and I was swayed by the post in Remote Recording so also thought it was to do with IR for reverb in mastering music.

If its just a school project where one might be happy with the first 20dB of T60 then a balloon might be OK.

Here are some other good papers.
http://www.fesb.hr/~mateljan/arta/pa...m-aaaa2003.pdf

and some excellent papers on balloons for RT measurement.
http://www.fesb.hr/~mateljan/arta/pa...c-aaaa2007.pdf
http://www.fesb.hr/~mateljan/arta/pa...t_aaaa2007.pdf

Note that the rooms are not 3000 seat LaScala and balloons of up to 160cm were used!

[fabiano83]

Thanks everybody for the replies! I am reading all the documents everyone linked :D
I think i probably won't do the measurements, the gun was a very good idea, but i think that i could have problems crossing the swiss/italian border with a firearm, i think that even if it is a starter gun it can't be introduced in italy... shame that i don't live in the US i might consider the balloon method if we visit smaller opera theaters...
the measurements are not really crucial to my dissertation, i will be discussing acoustics of theaters and i thought it could be cool to have my own measurements since i'll use a couple of theaters as examples, but i'll probably use the one that can be found on the internet, they are more professional
BTW i really love acoustics and i hope my passion could turn into a job... we'll see... next thing i'll buy is a starter gun, so i can go around town making measurements without getting disturbed