I think what you'd need to do is duplicate the track, pan one left, one right - use some light gating.

Wait - triple the track, sorry. One centre, one left, one right - then you'll have a heck of a job setting the parameters so they fit tightly and don't trigger badly.

So you want the gate to trigger on frequency AND dynamics. You could try Floorfish, it has a frequency knob on it, but don't know how accurate it is. Also Autogate VST, which has sliders for low pass filter and high pass filter - as Codemonkey suggests, you duplicate the track several times, set the filters differently on each track so that only the desired frequency is coming through on each, and then set the threshold for the gate. But I agree with him, it will be near impossible to have it sound natural. When you set the filters narrow enough to only capture a very narrow frequency, it will likely make it sound unnatural. But maybe you can go a bit wider and reach a decent compromise. So instead of using the gate to trigger panning, you're panning the entire duplicated track in advance and using the gate to just let what you want through over there on the left or the right.

Or what amounts to the same thing: first EQ the living hell out of each duplicated track so only the frequency you want is coming through, and then apply a different gate to each. Those plugins I mentioned just have the LPF and HPF built in, so it's easier, but maybe it's less effective than a dedicated and better quality EQ? Those plugins are both free though, so you could still use them just for the gate based on the dynamics. I say have at er, maybe you can make it work. Nothing to lose! Post the results. I'm curious to know how it goes.

Maybe try leaving one track with everything ungated in the center, so you hear everything through center, and then gated versions of the two panned trumpets panned far left and right, so that when the gated trumpet opens up far left for example, the little bit of other frequences you couldn't filter out over there blend in with the ones coming from the center. Although that would probably affect how wide the stereo spread sounds too - I wonder if it would make it sound like some of the sounds are moving around? Still, it would be a neat experiment to try.

Yes, this sort of panning is possible either using an external digital mixer or ITB with a moderately well-featured DAW.

BUT...have you actually heard what the results of this type of spatial sonic shuttling sound like? For an effect in the middle of one track, I could imagine it would be OK, but surely not over a full CD? You would certainly want to keep your fundamentals centred during panning of the trumpet overtones.

Assuming you have a stereo track to start with, what you could try is duplicating a mono mix of your stereo track to two mono tracks panned L and R, and then manually using volume envelopes to bring up the L and R when you identify the correct trumpets. See if you are comfortable with how that sounds before we go through blow-by-blow accounts of how you might set it up to be automatic.

To expand on this...

Rather than a gate, why not some clever envelope automation? The on/off nature of gates doesn't sound very good for this sort of thing, IMO.

My approach would be to draw on the waveform of the L (2nd) and R (3rd) channels. Do a nice, smooth envelope in and out whenever these trumpets sound.
Tedious? Yes, especially if you have a lot of notes. Much more natural sounding to me though. If the trumpets are louder enough than the drone, the notes should be pretty easy to find in the waveform.
Don't forget to apply a nuclear HPF to each just below the fundamental.

Is micing the drone and individual trumpets and panning them accordingly out of the question?

This topic really intrigues me for two reasons:
One, I had no idea a didg could produce that many overtones or trumpets!
Two, I see some interesting mix and DAW notions developing here.

Dvdhawk-
While I'm not fully aware of how you do the sort of thing on a didgerdoo that the OP is describing (I have trouble getting a good drone going), I think that would be somewhat analogous to trying to capture these individual notes on a horn.
I don't believe there's that much separation to work with.

Digeridan?

Have you tried any of the suggestions as of yet?

I've seen videos of guys doing the beatbox thing with them, but they've all looked like a single tube to me. The OP makes it sound almost like a bagpipes kinda thing with multiple trumpets. I clearly don't know much about the didgeridoo. I only ask about more mics because I can't wrap my brain around side-chaining gates to key on specific notes and auto-panning them. Interesting topic though. I will be watching the thread to see what Didgidan comes up with.

Thanks to everyone for your input. I'd like to somehow upload a sample for y'all to hear what's possible on a didge in terms of trumpets. It's on a didge that has six "official" trumpet notes and three additional trumpets that don't have a lot of backpressure but can be coaxed out of the instrument. It's all one pipe of course; a trumpet note is, technically speaking, one of the frequencies which is an intrinsic resonance of the pipe, that is, a frequency at which there is high acoustic impedance (or backpressure). The frequencies with high impedance are sounding as the additional harmonic overtones while the drone is being played, but they are also available to be sounded as trumpet notes by buzzing the lips much tighter than for the drone. You can only sound one trumpet note or the drone at a time, but each has its own set of background harmonic overtones.
A friend of mine has created software that uses transmission line modeling to model the interior form and predict the sound spectrum. (http://www.didgeridoo-physics.com/E/index.htm)
Using evolutionary algorithms, he "evolves" the shape necessary to get any desired set of sound and playing characteristics. The didge in my sound sample was created with this method to have multiple trumpet notes and exaggerated harmonics.

Anyway, to get to the point...

So far, I've tried envelopes on the waveform to pan, just to get a feel for what it would sound like if it were automatic. It sounds allright, but I'm looking for a way to make it automatic so that I can perform live in stereo.
The other thing I tried was tripling the track, and having one centered, one left, and one right. Then I used the equalizer on each panned track to eliminate all frequences except for a very narrow band around 135hz (2nd trumpet) on one and 209hz (3rd trumpet) on the other. I also filtered out those same frequencies from the centered track because it seems like I would need it to prevent phasing during trumpet notes. Is that right? Then I applied a gate to each. As predicted, it is very difficult to set it just right, and I still haven't got it. If the threshold is set too low, the gate opens at times when I am playing loudly but not hitting the appropriate trumpet note. So I turn the gate up and now it's not opening consistently enough, sometims the trumpet didn't come through. I've been using the spectrogram on Ableton to visualize the didge's frequency band, and can see that there really is a large difference between the decibel level of the trumpet note frequency when no trumpet is sounding and when it is. That is, 135hz does not go nearly as loud when it is a background harmonic to the drone as it does when punched as a trumpet. So, with such a distinct decibel difference, why should the gate be so finnicky?
Just turning the gate off, but with the same setup--three tracks, two panned and equalized--has worked best so far. The left and right tracks have to be turned down enough to blend in with the center track when the trumpet is not sounding, so it is not as exaggerated to right and left as I would like it.

Shouldn't I be able to set the gate so that it's threshold pertains to only one frequency, say 135hz, instead of responding to the whole spectrum? Or is this what I am already doing by using the equalizer to filter out all frequencies except for a narrow band? That is, is the gate still responding to the full spectrum band of what is coming in through the mic, or is it applied after the equalizer? Maybe this is where I should try out apstrong's suggestion with Floorfish or Autogate.

I sent the question of how to pan certain frequencies to Ableton's tech support too. Here's what I received; tell me what you think about it:

"you could probably create a rack which contains 2 device chains. Using a Utility device in each chain will allow you to split the stereo signal into 1 left and 1 right channel, so if you now load the gate into one of the chains, you can set it up to mute the signal as soon as it exceeds the threshold. As the other chain still passes an ungated signal, you will have the rack's output panned to 1 side. Not sure if this will work for you, but this could be a solution.

Another solution would be to use a 3rd party tool, such as Max/Msp. If you know how Max/Msp works, you could program yourself a patch which works like gate and then just triggers a MIDI CC as soon as the signal exceeds the threshold. Thus you could route the CC into Live's remote input and map it on the pan knob."

As to the first suggestion, wouldn't I still have the threshold problem? Maybe with the gate being flipped it would work out differently? I haven't yet learned how to use a rack, so I've got some homework to do if I'm going to try that suggestion.

On the second suggestion, I don't know what a patch is, nor a MIDI CC. More homework...

Personally I would avoid the gate altogether. They are finicky and it's difficult to get them to sound natural, especially if there is not a lot going on in the background and you don't have great separation between instruments. In my experience anyway. I would keep going down the path you mention here:

Just turning the gate off, but with the same setup--three tracks, two panned and equalized--has worked best so far. The left and right tracks have to be turned down enough to blend in with the center track when the trumpet is not sounding, so it is not as exaggerated to right and left as I would like it.

How about some compression, after the EQ in the chain, on the panned tracks to even out some of that discrepancy between the panned tracks and the center track. You can also do some subtractive EQ to the center track to remove more of the trumpets you want panned, so that when they kick in on the left and right, they blend less with what's coming down the middle and hence sound wider.

Shouldn't I be able to set the gate so that it's threshold pertains to only one frequency, say 135hz, instead of responding to the whole spectrum? Or is this what I am already doing by using the equalizer to filter out all frequencies except for a narrow band? That is, is the gate still responding to the full spectrum band of what is coming in through the mic, or is it applied after the equalizer? Maybe this is where I should try out apstrong's suggestion with Floorfish or Autogate.

It's the same thing when you EQ before you gate. The advantage of those plugins is that they have the EQ built in, and I believe they are programmed so the EQ is pre-gate (not sure). The gate should respond to only what's getting past the EQ filters. But I doubt you can set the EQ as precisely as you want - if you narrowed it so that only 135hz is coming through, you are getting rid of all the stuff around 135hz which contributes to the tone of the trumpet. I can't believe it generates a pure 135hz waveform and nothing else. This is why I like the approach you mentioned above, without a gate. You allow more than just 135hz through, but you narrow the frequency range to something reasonable that doesn't affect the tone too much and then compress it to get some consistency between center and panned tracks. Doing it this way will sacrifice some stereo width, there's no way around that using this method - the trumpet you want on the left is going to come through the middle somewhat and perhaps even through the right channel a little bit too, but perhaps the subtractive EQ to the center track (and maybe even adding a hint of subtractive EQ to the the right track) will help get it wide enough to satisfy you.

you could probably create a rack which contains 2 device chains. Using a Utility device in each chain will allow you to split the stereo signal into 1 left and 1 right channel, so if you now load the gate into one of the chains, you can set it up to mute the signal as soon as it exceeds the threshold. As the other chain still passes an ungated signal, you will have the rack's output panned to 1 side. Not sure if this will work for you, but this could be a solution.

Another solution would be to use a 3rd party tool, such as Max/Msp. If you know how Max/Msp works, you could program yourself a patch which works like gate and then just triggers a MIDI CC as soon as the signal exceeds the threshold. Thus you could route the CC into Live's remote input and map it on the pan knob."

As to the first suggestion, wouldn't I still have the threshold problem?

I don't fully understand suggestion #1, but my guess is that you'd still have the threshold problem with either suggestion. #$#(*$* gates. :)

ps. The second suggestion is basically to use a gate to trigger a midi control signal. A midi signal can tell a device to use a certain instrument, to play a certain note, but it can also tell a device to pan, or it can control the play, stop, and record functions on your DAW, or adjust the volume, etc.. It's just data, not sound, so it can be used to control all kinds of things. That's actually a pretty clever way to get a gate to control pan settings, but it's still a gate and you'll still have the threshold problem if you ask me.

And you can still do it live, you just have to route the same signal to three different channels and process each one differently. What I worry about now is latency if you're doing it through your computer. Worth trying though, and if the latency is unbearable, you can always do the same thing with hardware.

"I also filtered out those same frequencies from the centered track because it seems like I would need it to prevent phasing during trumpet notes. Is that right?"

No. So long as there's no delay involved, all you'd do is make the panning more effective (nothing in centre = purer when extra) - no phasing so long as the tracks don't move in time.

While applying EQ can cause some phase issues, the biggest issue is time, as Codemonkey pointed out. If you used two mics on the same source at different spacing, the source arriving at the mics at different times can create phasing. You've got one mic and one source, so I wouldn't be too worried about phasing.

My suggestion for using enveloping was more to model a gate but in a more gradual (and less absolute) manner.
For example, draw the waveform down to 50% (straight line). Then, slowly scoop up and down just before and after the trumpets. This would help keep the drone and other trumpets at a much lower volume, while allowing the individual trumpet to speak out more. I suggest 50% b/c really drastic changes in volume tend to sound unnatural. I do this sort of thing all the time when recording "live" and drums (or anything) get into the vocal mics more than I wanted.
The main problems with this method are two: One, it can be tedious; Two, it's drawn on the waveform, and the faders don't move. I like this b/c you can change the relationship in volume w/in a single track, and still use the fader to control the relationship in volume to other tracks. It's kind of like riding the gain knob post-recording. However, you couldn't do it live.

Apstrong's suggestions would much better serve you in a live setting.

Many thanks for the analysis and suggestions. Especially the tip to apply compression on the panned tracks. I think the gateless approach is going to work best. I'll write an update.

Soapfloats said:
"While applying EQ can cause some phase issues, the biggest issue is time, as Codemonkey pointed out. If you used two mics on the same source at different spacing, the source arriving at the mics at different times can create phasing. You've got one mic and one source, so I wouldn't be too worried about phasing."

Uhh.... well, actually, I've got two mics. While experimenting over the last few days, I've learned more about my two mics, a Shure Beta 58A and Rode NT1A. They each accentuate different parts of the sound spectrum, with the Beta getting the sharp, crisp high tones clearly and the Rode being more full-bodied. I'd like to record with both of them, not so much for stereo, as for timbre. Where they are placed, it doesn't sound like I'm getting phasing, but I'm no expert in recognizing it. The Beta is about 5" away from the bell end of the didge, pointing straight at the bell, and the Rode is back maybe 10", because, being a cardioid diaphragm type, it gets too bassy any closer. Would phasing be a problem when there is a 5" difference between the mics? If it is, maybe I could move the Rode closer to match the distance of the Beta and eliminate phasing, and then equalize out the added bass coming from the Rode mic. I was thinking of centering the tracks from both mics, and then making a duplicate of each mic's track, one for panning left, one for right.

Technically, for the track I'm working on now, I would be using three mics, if you count the subwoofer-mic I built out of a kickdrum and woofer. :) The subwoofer is in the open end of the drum, pointing at the drum head on the front side and the didge is played toward the drum head, which vibrates sympathetically with the didge. The subwoofer is routed in reverse to act as a mic instead of speaker, so it mics the low frequencies, (46hz(F#) is the fundamental frequency for both the didge and the drum.) I learned that drummers used this trick for adding bass to their recordings and thought I'd try it with a low hz didge. Couldn't believe how well it worked. Filtering the track from this mic has been straightforward and it complements the track from the smaller mics beautifully without saturating it with too much bass. This track is centered, of course and there are no phasing issues with it being on the low end of the spectrum.

3:1 rule ;)

Try 4" and 12" or 5" and 15". That'll minimise phase issues.

(Phase issues: you'll either hear bits of some notes missing, or reductions in the tone, or excess buildup of tone that overwhelms the non-built-up notes)
(If possible, flip the phase of one mic, you'll hear the difference that phase causes)

Didgeridan! Where are you?! How did it work out? You promised us an update. Don't make me release the attack squirrels.

update

Howdy y'alll,
Soon after our discussion I moved to a new city and have been very busy with some other projects. When I've unpacked my stuff, I'll get an mp3 copy of what I was able to come up with on Ableton posted onto my facebook profile. For now, if you're interested, check out my profile at facebook.com/didgeridan. There is already an mp3 posted on there; it is called "Drunk Horse". The version that's posted is mono, though. As soon as i can, I'll update it with the version containing panned trumpets.
Talk to you soon.
-Dan

Thanks, looking forward to the stereo version!

OK, as promised, I posted the new version of "Drunk Horse" with stereo effects on my facebook profile. facebook.com/didgeridan As you can hear, the stereo effect is only very slight. The first time the panned trumpets occur is about 32 seconds into the song--the lower trumpet of 135 hz is panned left. Then you'll hear the higher trumpet of 209 hz panned right for the first time at about 42 seconds. (Don't get confused, there are other trumpet notes that can be heard later in the song, but I only panned these two, since they occur most frequently.) The one on the left sounds satisfactory, but the one on the right doesn't show up as much, even though I processed it the same way.
Basically, I did what was suggested--not using a gate but just the equalizer. One duplicated track is stripped down to 135 hz and panned hard left and the other is stripped down to 209 and panned hard right. Also, on the main,, centered track I filtered out those same frequencies.

And me without a Facebook account. It makes me sad. Can you put it up somewhere else?

I uploaded it to a file hosting site. Here is the URL:

http://www.upload-mp3.com/files/75015_lah8i/Drunk%20Horse%20Final%20Cut1.mp3

That's very cool stuff! I hear the problem with the right trumpet. Unfortunately, I am out of ideas. You recorded this with two mics, correct? Random things I would try:

1. Panning the centre track off centre. Where? Who knows. Move it around and see what happens.
2. Increase volume of the right track by 3dB.
3. Move the right track forward (ahead of) the centre track by 10ms give or take. I'm totally grasping at straws now.
4. The Soapfloats method already mentioned: draw in a "gate", or essentially automate the fader on the right panned track so you're just allowing that note and no others to come through (and conversely on the centre track, cut that note out). It ought to work. Don't know how it will sound though! (in this method, you wouldn't eq the 209hz note out of the centre or vice versa for the right, just changing the volume for that one note - how much will depend on your ears before it sounds weird).
5. Repeat experiment with one mic. Then the other. Does it make a difference? I wonder, since you mention that the two mics accentuate different parts of the spectrum. Or use both and then bounce down to a mono file before you duplicate them left and right and do the EQ trick. What could it hurt? :) If you had two of the same mic, that would be interesting too, keeping the centre track stereo itself for a little bit of width, and then one of those also duplicated and panned left and one duplicated and panned right.

Sorry I don't have a good answer. Hopefully someone has some clever ideas that will get you the rest of the way there. You can also google "widen stereo field" or "widen stereo image" for more ideas.

That horse is totally drunk.

Yes, I recorded with two mics. I've already tried suggestions 1 and 2. In fact, in the mp3 I posted, the track from each mic is actually off center, one slightly right, one slightly left. Then the equalized trumpet note tracks are hard left and right. As to the third suggestion, (moving a track 10 ms forward) this wouldn't solve the issue of being able to recreate the effect live.
Right before I posted the mp3 I messed around some more with the gating idea. But the problem is with the threshold. It's either too low, allowing the gate to open when it shouldn't, or too high, not opening when it should.

Another thing: when I say that I panned one track completely right and one completely left, that includes one track from each mic, so, each trumpet actually has two tracks panned. So, there are two center tracks, two tracks panned left, and two tracks panned right. I had also tried just one track per trumpet at first. Either way doesn't really make much difference.

What I'm wondering now is whether there is a device that is like compression, only in reverse, which I could apply in the chain on a panned trumpet track, after equalization. So, as i've already done, I would have my 209 hz trumpet track hard right with equalizer applied to filter out everything except for 209 hz. Then, whenever a threshold decibel is reached (when the 209hz trumpet is sounded), instead of compressing the sound it would make it louder. Instead of a ratio of say, 3 to 1 meaning that for every increase of 3 decibels over the input threshold there would only be a 1 decibel increase in the output, it would be like a ratio of 1 to 3, so that for every 1 decibel increase over the input threshold, there is a corresponding 3 decibel output. Surely such a thing is already done and probably listed as a device in the software I'm using, I just don't know what it would be called. On the gate device, there is the option to "flip" it. I wonder if there is something similar on the compressor.
If I could do this, then I would be able to have the panned tracks turned down more than they currently are in the mp3 that I posted. One problem that I have with the mp3 as it currently stands, which I haven't mentioned until now, is that when the trumpets are not sounding, I can still hear the respective frequencies humming in the background a little too much.
I hope I'm making sense. Thanks for the input.

Expander

I did a little google search and found out the name of what I'm talking about: an expander.

http://www.mediacollege.com/audio/processing/expansion/

I suppose you probably already knew all this. I guess the only difference in the expansion device spoken of at that link and what I was imagining is that the threshold works in the opposite way. "Any signal below the threshold is expanded downwards by the specified ratio." Still, the net effect should be the same, allowing me to crank the panned trumpet track loud enough to hear the trumpet when it sounds while not having to hear the steady background hum at 209hz.

Upwards expansion to be specific...

http://www.wavosaur.com/vst/noise-gate-plugins.php
See the 2nd on the list.

You could gate a little (use the range control to limit the gating) and apply the expansion as well.
There are plenty more plugins but this works great for what I've used it for and is free.
(It has a filter on the detector too).

Thanks Codemonkey.
...............................................................................................................
http://www.wavosaur.com/vst/noise-gate-plugins.php
See the 2nd on the list.
...............................................................................................................

It looks like that one is only available for windows and I'm using a Mac. So I tried Floorfish instead. But for some reason, the VST is not showing up in Ableton's plug-in device browser. I made sure to put the VST in the right folder (library/audio/plug-ins/VST) and to select "Use VST Plug-in System Folders" in Ableton, but it won't recognize it. I'm pretty sure Floorfish is supposed to be compatible with Live. This is, however, the first time I've ever tried to use a plug-in, so maybe I'm missing something. I emailed Ableton's tech support and am waiting for a response. I might just upgrade to Live 8 since it has an expander and some other things I want.

You might want to consider trying this in Reaper? Reaper has built-in parameter modulation features that might do the trick, without any 3rd party plugins at all: Reaper allows you to modulate any automatable parameter for any plugin using the dynamics of another channel. The demo is unrestricted... ;-)

Thanks. I'll post an updated mp3 after messing around with an expander.