KVR Audio

guitarzan · Post by **guitarzan** » Wed Apr 12, 2006 5:20 pm

It sounds like a great idea to me because it could be built on any scale - little laptop Leslies...what a world it will be!

AdmiralQuality · Post by **AdmiralQuality** » Wed Apr 12, 2006 5:33 pm

Guys, it can't work. For all the reasons mentioned above. Sorry.

You can't make it "_real_" because you can't calculate the doppler or phase shift unless it's _FAKE_. (The microphone or human head has to be part of the model, without that there's no saying in which direction the doppler should shift... among a 100 other problems)

Spatialization only works because the speakers are AROUND you. Projecting from 4 (or 2) unmoving, central speakers out into the world won't indroduce any doppler, and only a bit of phase shift (which would come off the surrounding walls in your room, and only amount to a crossfade between 4 static phase shifts anyway).

Sorry, don't mean to rain on this parade... maybe it'll be SOMETHING. But it won't sound like a Leslie (unless maybe you run around it really fast.)

Heheh.. which reminds me of this demo video:
http://www.rolandus.com/products/produc ... arentId=57 (SKIP the intro in the FLASH and then click VIDEO to see the demo dude. He does a human Leslie at the end. But at least his actually spins.)

multree · Post by **multree** » Wed Apr 12, 2006 5:41 pm

something here

michu · Post by **michu** » Wed Apr 12, 2006 6:30 pm

quinto,
to get get doppler effect you'll most probably need to include some modulating delay lines. just morphing between impulses may not cut it...
the idea of having a box that you can stand behind and have a feeling of standing behind real leslie is quite interesting, tho to be viable live it would IMO have to have the DSP part on board. (given that you'd need at least a rackmount pc or a laptop and a multi IO audio interface to run your solution, maybe you should investigate some barebone pc running linux from bootROM or something)
for home/studio use the 2 speaker solution that Admiral Quality mentioned should be more than enough. you don't care about the sonic radiation into the wall

the idea of making leslie emulation by using impulses of leslie with speakers at various angles seems very promising tho

pj geerlings · Post by **pj geerlings** » Wed Apr 12, 2006 6:55 pm

J.O.Smith's approach to this challenge

It's worth a look - I had to study these pages quite a bit before coming up with my own solution ...

peace,
pj

guitarzan · Post by **guitarzan** » Wed Apr 12, 2006 7:12 pm

AdmiralQuality wrote:Guys, it can't work. For all the reasons mentioned above. Sorry.

You can't make it "_real_" because you can't calculate the doppler or phase shift unless it's _FAKE_. (The microphone or human head has to be part of the model, without that there's no saying in which direction the doppler should shift... among a 100 other problems)

Spatialization only works because the speakers are AROUND you. Projecting from 4 (or 2) unmoving, central speakers out into the world won't indroduce any doppler, and only a bit of phase shift (which would come off the surrounding walls in your room, and only amount to a crossfade between 4 static phase shifts anyway).

Sorry, don't mean to rain on this parade... maybe it'll be SOMETHING. But it won't sound like a Leslie (unless maybe you run around it really fast.)

Heheh.. which reminds me of this demo video:
http://www.rolandus.com/products/produc ... arentId=57 (SKIP the intro in the FLASH and then click VIDEO to see the demo dude. He does a human Leslie at the end. But at least his actually spins.)

The speakers are centrally locatad pointing outward - like toward the compass points - not surrounding you like stereo. Then the signal goes from speaker to speaker blasting out at you around a central point, not from around your head.

AdmiralQuality · Post by **AdmiralQuality** » Wed Apr 12, 2006 7:30 pm

guitarzan wrote:The speakers are centrally locatad pointing outward - like toward the compass points - not surrounding you like stereo. Then the signal goes from speaker to speaker blasting out at you around a central point, not from around your head.

I know. But you only hear sound coming at you. It's directional. Saying "blasting out at you around a central point" is inherently oxymoronic. Unless it's coming from inside your own head.

So if you have n speakers, the best and only place to point any of them is at your head. (IMHO! obviously it depends on the room design too.)

Now, that said, I'm willing to back off on my stance a bit because if you do everything else that a typical Leslie sim does -- which is what people are starting to suggest -- then it will probably sound Leslie-like. I just don't think the speakers pointing out arrangement will add anything to it. IMHO you'd be better off with 2 speakers pointed at your 2 ears.

But I'm willing to be surprised and don't mean to discourage you totally and will be happy to help test it.

(Thanks for the link PJ!)

guitarzan · Post by **guitarzan** » Wed Apr 12, 2006 7:36 pm

It would be a performance thing - not a recording tool, unless you mic'ed it up or something. Like a real Leslie. It's the opposite of surround sound. The speakers are grouped together in the center of the room (for example) - back to back pointing out toward the walls. The sound goes from speaker to speaker - sounding much like it would if a single speaker were spinning in the middle of the room

quintosardo · Post by **quintosardo** » Wed Apr 12, 2006 9:12 pm

Surely I did some confusion speaking about emulation, IRs and so on...

I would focus on the main part:

replace a rotating horn with four fixed loudspeakers. Each louspeaker pointing from the center to one 90° section. On a cross for 360°.
Then calculate vector projection to fade from only one loudspeaker to only the adjacent one.
I.e.:
speaker1: 0
speaker2: 0
speaker3: A sin(angle)
speaker4: A cos(angle)

where angle goes from 0° to 90° and so on

Nothing more than this. With the hypotesis that we can set good relative phases among speakers, shouldn't this be like a rotating horn?

Please don't get me wrong, consider that any criticism here is _more_ than welcome

From this point everything could be added: IRs, external case, original variations, etc...

Two loudspeakers are enough for emulation, but for this vectorial combination you need 4 orthogonal ones...

And yes... if ever this will seem like an interesting thing I'll build it

Kiwburger · Post by **Kiwburger** » Wed Apr 12, 2006 11:03 pm

I've often thought about exactly this sort of thing. I've owned real Hammonds and Leslies before, sold them due to the excess noise from the motors. I think the only solution is to build a quiet motor version.

Simply panning between speakers will have zero doppler effect, and just is not worth it. You need to ADD VELOCITY to the soundwave with motion.

I find the panning aspect of virtual leslies very disorienting, and usually set them to zero width (mono). You have to consider how real Leslies are mic'd normally. It's not actually common to use a stereo pair and have the sound pan wildly across the stage. It's more common to have two mics, and maybe pan them for a little spread between horn and rotor.

The sound of a real leslie in a room is amazing - but when recorded, it's essentially a mono sound. Or maybe spread a little between highs and lows - but not a auto pann effect. That would be sick making.

guitarzan · Post by **guitarzan** » Wed Apr 12, 2006 11:29 pm

It really seems like it would work to me - the sound would actually spin in the room!

It's not virtual, the speakers are real - not in the corners of the room pointed at you - they are grouped together back to back to back to back in, say, the center of the room right in front of you - a speaker pointing TOWARD each of the the four walls. To record it would require a mic, just like a real Leslie.

First the sound would come right out at you, then pan to come out of a speaker pointing TOWARD the wall to your LEFT, then it would pan to come out of a speaker pointing AWAY from you, toward the wall in front of you, then pan to a speaker pointing TOWARD the wall to your RIGHT. Then back to directly at you ...etc. I don't know if my direction of rotation is correct, I'm just trying to make a point.

Dang it, ya got a picture quintosardo?

Hink · Post by **Hink** » Wed Apr 12, 2006 11:41 pm

guitarzan wrote:It really seems like it would work to me - the sound would actually spin in the room!

It's not virtual, the speakers are real - not in the corners of the room pointed at you - they are grouped together back to back to back to back in, say, the center of the room right in front of you - a speaker pointing TOWARD each of the the four walls. To record it would require a mic, just like a real Leslie.

First the sound would come right out at you, then pan to come out of a speaker pointing TOWARD the wall to your LEFT, then it would pan to come out of a speaker pointing AWAY from you, toward the wall in front of you, then pan to a speaker pointing TOWARD the wall to your RIGHT. I don't know if my direction of rotation is correct, I'm just trying to make a point.

Dang it, ya got a picture quintosardo?

with reflections and all I think your method might work better if instead of the speakers facing 90 degrees from center they should be more like 45 degrees off center...though that's assuming the speaker is not in the center of a room but considerably closer to one wall...

still I'm not sure it would be all that great...

guitarzan · Post by **guitarzan** » Wed Apr 12, 2006 11:55 pm

Hey Hink! It's not my idea - it's quintosardo's from SKnote. It just seems like everybody is picturing this as a normal virtual rotator plugin, and it's not. It's a physical array of speakers pointed out away from a common axis at 90 degree angles that the signal then pans around. Seems to me you'd have genuine spin. It would be more a live performance thing in my mind - not a virtual Leslie plugin like already exists. I'll just wait and see I guess.

Kiwburger · Post by **Kiwburger** » Thu Apr 13, 2006 12:02 am

It's simply sound panning - which is nothing like the beautiful Leslie chorus effect. There would be zero pitch bending, therefore zero chorusing. All you would have is a stepping volume attenuation, with stepping phase issues. If the goal is to have something like a home organ consol, it might be interesting - if disturbing. But if the goal is to then re-record it, and sound somewhat like a Leslie, it would be complete failure.

A better compromise might be to use the VST to apply the missing doppler shift, and send this to stationary speakers.

Kiwburger · Post by **Kiwburger** » Thu Apr 13, 2006 12:07 am

It's just a whole lot cheaper and better to have moving reflector. I think a stepper motor could do this, and if the stepper motor was under VST control, that would be brilliant. I've suggested this idea on another forum a long time ago, and got zero interest and much heckling.

Would you use this _real_ Leslie? AUDIO DEMO