Subject: how would I calculate and apply a compensation curve for parabolic dish?

Hello,

I use a parabolic dish for capturing wildlife audio, mainly birds. Even though the dish is approximately 30 inches in diameter, it still
emphasizes upper frequencies more than lower.

I would like to determine the frequency response of the dish and then
apply a compensation curve in software to any audio that was recorded
with the dish. How might I go about doing this?

My initial thought was to have either pink noise or sweeping audio
frequencies being emitted from a speaker, then capture the resulting
spectra with first the audio recorder by itself and then the audio
recorder/ dish combination. Then I would use software to see the curves
and maybe use a VST plug-in to match the parabolic system audio to the recorder by itself. Well, here I could either try a software match or
change the curves manually.

Your ideas for this task would be welcome and thanks in advance.

JB

---
* Origin: rbb pygate - the fidonet nntp junction (2:221/10.0)

Subject: Re: how would I calculate and apply a compensation curve for parabolic dish?

JB Wells <jbw194@anytime.net> wrote:

I use a parabolic dish for capturing wildlife audio, mainly birds. Even >though the dish is approximately 30 inches in diameter, it still
emphasizes upper frequencies more than lower.

I would like to determine the frequency response of the dish and then
apply a compensation curve in software to any audio that was recorded
with the dish. How might I go about doing this?

The problem is that the frequency response is going to change dramatically
when you move it a degree or two, and it is going to be full of giant narrowband resonances.

Try it for yourself... set up a speaker outdoors with some software that
allows you to do simple swept-sine tests.... maybe something like Room
EQ Wizard. Do sweeps from 500 Hz up to 10Khz with the microphone maybe
25 to 50 feet away and watch the plot on the screen. It's going to be
super ugly and the ugliness changes if you even touch the position a
tiny bit. You'll find peaks and dips that are much narrower than a single note, and that move around a lot.

Don't worry about ground reflections... normally if you do a microphone
test like this the reflections from the ground cause a problematic midrange resonance and you need to raise the speaker up... but in this case that
is going to be totally dwarfed by the ringing of the reflector itself.

My initial thought was to have either pink noise or sweeping audio >frequencies being emitted from a speaker, then capture the resulting
spectra with first the audio recorder by itself and then the audio
recorder/ dish combination. Then I would use software to see the curves
and maybe use a VST plug-in to match the parabolic system audio to the >recorder by itself. Well, here I could either try a software match or
change the curves manually.

Swept sine is the easy way to do it. These days most folks are using
MLSSA or similar impulses and then deriving the frequency response from
the impulse response, which can give better results at low frequencies
faster and can let you gate out room problems. But you don't care about
low frequencies or room problems, so just go with a simple swept sine.

But once you see what the curves look like and how unstable they are,
you're going to realize that what you want is impossible. You might
be able to notch out one or two modes in order to make for a slightly
more natural voice sound but even that is problematic.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

---
* Origin: rbb pygate - the fidonet nntp junction (2:221/10.0)

Subject: Re: how would I calculate and apply a compensation curve for parabolic dish?

On 4/12/26 9:38 AM, Scott Dorsey wrote:

But once you see what the curves look like and how unstable they are,
you're going to realize that what you want is impossible. You might
be able to notch out one or two modes in order to make for a slightly
more natural voice sound but even that is problematic.
--scott

Thanks. Well, I had already attempted several curve corrections this
past week but not all that successful, probably because of what you
mention.

I didn't use any generators, but there is a bird that lands on the same
tree nearly every day and calls for about 3 minutes. Honing in on it
from about ten feet away with the dish was easy; trying to get a
recorder up there near the bird and pick up its calls with decent
amplitude was another matter, but finally had recent success.

It was a Northern Cardinal and they emit several calls that aren't
complicated and easy to remember by ear. I picked two matching calls,
one with and the other without the dish, and then attempted a frequency correction using Nova GE. I had success one time out of three attempts.
Unless the dish is oriented exactly the same way from the same
distance, along with the recorder, I'm never going to be able to devise
the same matching curve that can be applied each time. I think for the
very reason you cited above.

I wonder how the supposed pros do it? I thought maybe I'd find a
documented process somewhere but to no avail. I read mention of the
high frequency boost from the dishes, but when I look at bird sites
where professional birders post their audio files, although some mention
the use of a gentle high pass filter to cut traffic nearby (I do as
well), no one mentions using a correction curve on their audio.

One process that may help me somewhat is something Nova GE has called a "dynamic deresonate" which tries to reduce resonances as they appear,
which seems to sound more natural than if I try using the "static"
version of the filter. And it definitely reduces parts of the spectrum
from about 3-8 Khz up to around 20 db.



---
* Origin: rbb pygate - the fidonet nntp junction (2:221/10.0)

Subject: Re: how would I calculate and apply a compensation curve for parabolic dish?

On 13/04/2026 2:59 am, JB Wells wrote:

On 4/12/26 9:38 AM, Scott Dorsey wrote:

But once you see what the curves look like and how unstable they are,
you're going to realize that what you want is impossible.ÿ You might
be able to notch out one or two modes in order to make for a slightly
more natural voice sound but even that is problematic.
--scott

Thanks.ÿ Well, I had already attempted several curve corrections this
past week but not all that successful, probably because of what you
mention.

I didn't use any generators, but there is a bird that lands on the same
tree nearly every day and calls for about 3 minutes.ÿ Honing in on it
from about ten feet away with the dish was easy; trying to get a
recorder up there near the bird and pick up its calls with decent
amplitude was another matter, but finally had recent success.

It was a Northern Cardinal and they emit several calls that aren't complicated and easy to remember by ear.ÿ I picked two matching calls,
one with and the other without the dish, and then attempted a frequency correction using Nova GE.ÿ I had success one time out of three attempts.
ÿUnless the dish is oriented exactly the same way from the same
distance, along with the recorder, I'm never going to be able to devise
the same matching curve that can be applied each time.ÿ I think for the
very reason you cited above.

I wonder how the supposed pros do it?ÿ I thought maybe I'd find a
documented process somewhere but to no avail.ÿ I read mention of the
high frequency boost from the dishes, but when I look at bird sites
where professional birders post their audio files, although some mention
the use of a gentle high pass filter to cut traffic nearby (I do as
well), no one mentions using a correction curve on their audio.

One process that may help me somewhat is something Nova GE has called a "dynamic deresonate" which tries to reduce resonances as they appear,
which seems to sound more natural than if I try using the "static"
version of the filter.ÿ And it definitely reduces parts of the spectrum
from about 3-8 Khz up to around 20 db.

Wireless mic (omni lavalier ?) as close to the likely source.
--
geoff

---
* Origin: rbb pygate - the fidonet nntp junction (2:221/10.0)