Consul
Well-known member
I guess it was too much to hope that this would turn into an interesting DSP discussion.
:razz:
:razz:
Amp modeler for tweak-heads
- Thread starter Emperor-TK
- Start date
Help Support GroupDIY Audio Forum:
While I worked there Hartley had one or more planes. A nice King Air turbo prop for short hops and a I think a G-II (later a G-IV) that made visits to Corby or Frankfurt less painful (if we ignore that one time we returned back into the country via Miami :roll: from parts south). So I find it a little odd that somebody would brag they flew Hartley somewhere.
I could imagine some symbolic laying on of hands by Hartley, if he was romancing Eddie to participate in a larger commercial venture. Hartley was surely capable of tweaking on a Marshall (or whatever), if suitably motivated. IMO it would take more than a plane ticket and a back stage pass.
I guess it depends on who is telling the story, to whom.
JR
I could imagine some symbolic laying on of hands by Hartley, if he was romancing Eddie to participate in a larger commercial venture. Hartley was surely capable of tweaking on a Marshall (or whatever), if suitably motivated. IMO it would take more than a plane ticket and a back stage pass.
I guess it depends on who is telling the story, to whom.
JR
...I'm still reading about masses, springs, and dashpots.
I was thinking along the lines of finding the place programatically where SPICE determines an output graph and trying to import it into another function (recognized as a function by an audio render program; plugin)
I guess that would include replacing the SPICE signal generator with a representation of your own wave form as well.
I thought a good place to start would be to try to find a simple open source circuit modeler and insert either references to an audio/data stream or a standalone file for translation to WAV to be called by another program or plugin shell.
The summary of my initial idea was to use the equations in Circuitmaker to render audio through the circuit, but I lack the software toolbox to decompile it. Any suggestions? Simple is good right now.
I'm thinking Microsoft SDK stuff here, it's free.
It's just a hobby for me, but I do appreciate the insight of those here who are more knowledgable than myself.
I was thinking along the lines of finding the place programatically where SPICE determines an output graph and trying to import it into another function (recognized as a function by an audio render program; plugin)
I guess that would include replacing the SPICE signal generator with a representation of your own wave form as well.
I thought a good place to start would be to try to find a simple open source circuit modeler and insert either references to an audio/data stream or a standalone file for translation to WAV to be called by another program or plugin shell.
The summary of my initial idea was to use the equations in Circuitmaker to render audio through the circuit, but I lack the software toolbox to decompile it. Any suggestions? Simple is good right now.
I'm thinking Microsoft SDK stuff here, it's free.
It's just a hobby for me, but I do appreciate the insight of those here who are more knowledgable than myself.
I have more questions than answers about DSP (currently teaching myself to perform FFT).
One question when trying to use DSP to model real things, there are differences between FIR and IIR filters where they don't both deliver the same phase response as analog RC networks.
This is either a problem, or a new solution for something...
JR
One question when trying to use DSP to model real things, there are differences between FIR and IIR filters where they don't both deliver the same phase response as analog RC networks.
This is either a problem, or a new solution for something...
JR
Consul
Well-known member
I once dreamed of the idea of doing a DSP real-time circuit simulator, which could emulate both linear and non-linear properties. Now, though, I've thought more about what kind of sounds can be made in DSP that cannot (at least easily) be done in the analog world. The trick will be to spend time finding algorithms that sound pleasing to our ears in their own way. I'd also note that it took early analog designers a while to find ways to design circuits with pleasing sound. DSP is still quite new in the grand scheme of things.
I'm not just talking about guitar amp sims, here, though that is part of it. Also software synthesizers, and methods of processing recordings into new and pleasing sample sets for samplers.
Here's something fun as far as that latter point is concerned: http://taps.cs.princeton.edu/
Here's a video of it in action: http://soundlab.cs.princeton.edu/listen/taps/tapestrea.mov (67MB)
My thinking is to use something like Tapastrea to post-process samples of just about anything (flicking a lampshade, for example) into new and pleasing samples for musical use.
I'm not just talking about guitar amp sims, here, though that is part of it. Also software synthesizers, and methods of processing recordings into new and pleasing sample sets for samplers.
Here's something fun as far as that latter point is concerned: http://taps.cs.princeton.edu/
Here's a video of it in action: http://soundlab.cs.princeton.edu/listen/taps/tapestrea.mov (67MB)
My thinking is to use something like Tapastrea to post-process samples of just about anything (flicking a lampshade, for example) into new and pleasing samples for musical use.
Consul
Well-known member
Something I've been thinking about, to carry on from my previous post above, has to do with "atomic units" for DSP. These would be analogous (no pun intended) to resistors, capacitors, and inductors in the electronics world, in the sense of all complex structures breaking down into collections of those atoms.
In DSP, your atomic units are integrators, delay lines, and multipliers. (One might also include memoryless waveshapers into that mix, for the non-linear stuff, but for now, we'll leave those out.) The problem here is that we're still, for the most part, tied to converting analog representations of circuits into differential equations (bilinear transform) so they can be implemented in software.
The approach by Zavalishin above, performing the bilinear transform on the integrator first and making that an atomic unit, marks the start of a system wherein different DSP structures can be built without needing to know about their analog counterparts. Just like in the analog world, different functions will have different topologies, and those topologies can be strung together in series and parallel to create complete "devices."
Once that barrier is broken, finding new and different ideas for sound making/mangling devices takes on a different approach.
The main problem I'm having right now, is that even though I understand the paper at an "overview" level, I'm still stuck on what he's telling me about how to actually implement the thing.
In DSP, your atomic units are integrators, delay lines, and multipliers. (One might also include memoryless waveshapers into that mix, for the non-linear stuff, but for now, we'll leave those out.) The problem here is that we're still, for the most part, tied to converting analog representations of circuits into differential equations (bilinear transform) so they can be implemented in software.
The approach by Zavalishin above, performing the bilinear transform on the integrator first and making that an atomic unit, marks the start of a system wherein different DSP structures can be built without needing to know about their analog counterparts. Just like in the analog world, different functions will have different topologies, and those topologies can be strung together in series and parallel to create complete "devices."
Once that barrier is broken, finding new and different ideas for sound making/mangling devices takes on a different approach.
The main problem I'm having right now, is that even though I understand the paper at an "overview" level, I'm still stuck on what he's telling me about how to actually implement the thing.
> I guess it was too much to hope that this would turn into an interesting DSP discussion.
You got that right.
> I could imagine some symbolic laying on of hands by Hartley,.... IMO it would take more than a plane ticket and a back stage pass.
As you say, who is telling the story? Who is supposed to be impressed? And great men don't rest on their reputation, they actively create it. Who flew who is not really critical to the story, just flavor. Between them, Eddie and Hartley and all their staff built a beast so great that it survived their later parting-of-ways, and still enhances both reputations.
> ...I'm still reading about masses, springs, and dashpots.
I was startled to see that dropped-in as if it were new. It is the root fiction of practical engineering on dynamic systems. You can see EVERYTHING as masses, springs, dashpots, and a few other basic-bits. Everything from atoms to universes is built this way (or we can pretend it is, and get useful answers). Even electrons in bottles. Oh, some springs obey Hooke and some obey Newton.... plug-in the formula for whichever "spring" you have.
> In DSP, your atomic units are integrators, delay lines, and multipliers
Well, OK, there are surely several ultimately equivalent ways to skin, slice, and dice the cat. And it varies depending if you want rump-roast, drumsticks, or burgers. This list of "atoms" favors serial streams.
And given that "anything" can be modeled in such terms, you can model a proposed or impractical plan on a sorta-universal machine.
Such a machine, once called Differential Analyzer, currently called DSP or just PC, may be limited or general. For most "interesting" problems they are limited because they can not compute all the springs and masses of interest. Can you model every electron in a tube amp? The computation is much bigger than the thing being computed. The first trick is figuring how few springs/masses we can get away with and still get interesting answers. A related trick is to efficiently organize, simplify, and sequence the computations. (There are practical difficulties with massively parallel computations.)
The Revalver can give an excellent approximation to a 10KHz 13-tube amp, but not very-good real-time approximation on a 1,000,000KHz 100,000,000 transistor computer. You don't build a Pentium to avoid building a DeLuxe or a 5150. But now that the demand for email and Quake and WoW has given us Pentiums at low-low marginal cost, what the heck.
> DSP structures can be built without needing to know about their analog counterparts.
To begin with, the wealth of common-sense experience and rigorous analysis of analog systems is the way to go.
And of course the throwing-off of our analog blinders must be a next step along enlightenment, though one I'm not ready for.
> massively DSP based sound systems (Media Matrix) replacing racks full of black boxes in large installs with virtual gear.
Yeah, that too. Some slick-suit contractor sold the school a PA rack which is one black box with a 400-page manual. Where's the crossover? The limiter? The EQ sliders? Can I get a delay here-here-here....? YES! But not the first time standing on graduation field an hour before showtime. I usta be able to whomp-up a LM324 crossover an hour before show; I'm not eager to re-learn totally GUI virtual sound processing. (But when the installation is FIXED, and non-simple, this GUI/DSP sure could be better. And the school/church market is significant to a rock+roll manufacturer.)
> post-process samples of just about anything (flicking a lampshade, for example) into new and pleasing samples for musical use.
One of the most fascinating and musical things I ever heard from an analog synthesizer studio was the day the synth and most of the other gear was dead. The student who'd booked the time, and was frustrated, picked up a big rubber band and plucked it. He found a cheap portable tape machine, he found a razor, you listen to the result and it is pretty catchy but what the HELL is it???
> stuck on what he's telling me about how to actually implement the thing
Alex Graham Bell knew some about sound and wires. But if you asked him to set up and run Hot Sushi's concert sound system, he'd have to learn and internalize a lot of stuff. We can't even see all the possibilities, let alone intuit them.
> a lone amateur built the Ark. A large group of professionals built the Titanic
The Titanic stayed afloat longer than the Ark.
You got that right.
> I could imagine some symbolic laying on of hands by Hartley,.... IMO it would take more than a plane ticket and a back stage pass.
As you say, who is telling the story? Who is supposed to be impressed? And great men don't rest on their reputation, they actively create it. Who flew who is not really critical to the story, just flavor. Between them, Eddie and Hartley and all their staff built a beast so great that it survived their later parting-of-ways, and still enhances both reputations.
> ...I'm still reading about masses, springs, and dashpots.
I was startled to see that dropped-in as if it were new. It is the root fiction of practical engineering on dynamic systems. You can see EVERYTHING as masses, springs, dashpots, and a few other basic-bits. Everything from atoms to universes is built this way (or we can pretend it is, and get useful answers). Even electrons in bottles. Oh, some springs obey Hooke and some obey Newton.... plug-in the formula for whichever "spring" you have.
> In DSP, your atomic units are integrators, delay lines, and multipliers
Well, OK, there are surely several ultimately equivalent ways to skin, slice, and dice the cat. And it varies depending if you want rump-roast, drumsticks, or burgers. This list of "atoms" favors serial streams.
And given that "anything" can be modeled in such terms, you can model a proposed or impractical plan on a sorta-universal machine.
Such a machine, once called Differential Analyzer, currently called DSP or just PC, may be limited or general. For most "interesting" problems they are limited because they can not compute all the springs and masses of interest. Can you model every electron in a tube amp? The computation is much bigger than the thing being computed. The first trick is figuring how few springs/masses we can get away with and still get interesting answers. A related trick is to efficiently organize, simplify, and sequence the computations. (There are practical difficulties with massively parallel computations.)
The Revalver can give an excellent approximation to a 10KHz 13-tube amp, but not very-good real-time approximation on a 1,000,000KHz 100,000,000 transistor computer. You don't build a Pentium to avoid building a DeLuxe or a 5150. But now that the demand for email and Quake and WoW has given us Pentiums at low-low marginal cost, what the heck.
> DSP structures can be built without needing to know about their analog counterparts.
To begin with, the wealth of common-sense experience and rigorous analysis of analog systems is the way to go.
And of course the throwing-off of our analog blinders must be a next step along enlightenment, though one I'm not ready for.
> massively DSP based sound systems (Media Matrix) replacing racks full of black boxes in large installs with virtual gear.
Yeah, that too. Some slick-suit contractor sold the school a PA rack which is one black box with a 400-page manual. Where's the crossover? The limiter? The EQ sliders? Can I get a delay here-here-here....? YES! But not the first time standing on graduation field an hour before showtime. I usta be able to whomp-up a LM324 crossover an hour before show; I'm not eager to re-learn totally GUI virtual sound processing. (But when the installation is FIXED, and non-simple, this GUI/DSP sure could be better. And the school/church market is significant to a rock+roll manufacturer.)
> post-process samples of just about anything (flicking a lampshade, for example) into new and pleasing samples for musical use.
One of the most fascinating and musical things I ever heard from an analog synthesizer studio was the day the synth and most of the other gear was dead. The student who'd booked the time, and was frustrated, picked up a big rubber band and plucked it. He found a cheap portable tape machine, he found a razor, you listen to the result and it is pretty catchy but what the HELL is it???
> stuck on what he's telling me about how to actually implement the thing
Alex Graham Bell knew some about sound and wires. But if you asked him to set up and run Hot Sushi's concert sound system, he'd have to learn and internalize a lot of stuff. We can't even see all the possibilities, let alone intuit them.
> a lone amateur built the Ark. A large group of professionals built the Titanic
The Titanic stayed afloat longer than the Ark.
Consul
Well-known member
[quote author="PRR"]> I guess it was too much to hope that this would turn into an interesting DSP discussion.
You got that right.[/quote]
Well, at least you're honest.
And it also seems you understand what I'm trying to say. The sooner we can break DSP away from its analog chains, the sooner we can turn DSP and the world of digital audio into something useful and interesting, rather than a pale imitation of the analog world.
Musique concrète. That's what's been making me think about all of this in the first place. Well, that, and that new Omnisphere softsynth, and their supposed "secret sample processing" that they do, which I have some suspicions about.
But the Ark actually fulfilled its first and only mission successfully. :wink:
That is, of course, you believe in its existence. Dave Barry's quote is only meant to motivate people out of a low sense of self-worth. I need the constant reminder.
You got that right.[/quote]
Well, at least you're honest.
And it also seems you understand what I'm trying to say. The sooner we can break DSP away from its analog chains, the sooner we can turn DSP and the world of digital audio into something useful and interesting, rather than a pale imitation of the analog world.
Musique concrète. That's what's been making me think about all of this in the first place. Well, that, and that new Omnisphere softsynth, and their supposed "secret sample processing" that they do, which I have some suspicions about.
But the Ark actually fulfilled its first and only mission successfully. :wink:
That is, of course, you believe in its existence. Dave Barry's quote is only meant to motivate people out of a low sense of self-worth. I need the constant reminder.
Butterylicious
Well-known member
I was expecting a bikini girl holding an amp. Imagine my disappointment. I was OK trading my tape deck and mixer in for a keyboard and mouse, but now I'm supposed to trade my amps in for a laptop. I know it's a recording thang, but I can't help picture a guitar player on stage with a wall of laptops. :grin:
keefaz
Well-known member
[quote author="Butterylicious"]but I can't help picture a guitar player on stage with a wall of laptops. :grin:[/quote]
At this point, I guess it will be more manageable to keep the analog amps and model the guitar player himself :wink:
At this point, I guess it will be more manageable to keep the analog amps and model the guitar player himself :wink:
okgb
Well-known member
How about this , It ain't quite there yet
but the more we get hammered with MP3's and the like
the closer modeling gets to being acceptable .
I can't afford an analoge synth either but they suffer more
where it doesn't need to be real time performance these days
In the Physical realm , modeling will never be loud
more of a craft and less of a performance
The modelers need a temperature control for heating the tubes over the course of an evening ,
that can sound different , and a spurious noise generator
I find them good for quick recording and not worrying about
isolation while still getting acceptable tones , last i compared
i felt the TDM version sounded better than the hardware
[ ampfarm ] i own two hardware versions
Fun marketing , the detail on the peavey ? can you get the continuous
reaction from modeling ? some parts of that boarder on the same
analoge / digital debate , a good tool no doubt and follows the rule
" It's the player , not the gear " there's a great R & B guitarist from
Regina , Sask. Jack Semple , used an SS peavey amp for a long time
, the the rock guys in their desperation seem more suseptable to
marketing
but the more we get hammered with MP3's and the like
the closer modeling gets to being acceptable .
I can't afford an analoge synth either but they suffer more
where it doesn't need to be real time performance these days
In the Physical realm , modeling will never be loud
more of a craft and less of a performance
The modelers need a temperature control for heating the tubes over the course of an evening ,
that can sound different , and a spurious noise generator
I find them good for quick recording and not worrying about
isolation while still getting acceptable tones , last i compared
i felt the TDM version sounded better than the hardware
[ ampfarm ] i own two hardware versions
Fun marketing , the detail on the peavey ? can you get the continuous
reaction from modeling ? some parts of that boarder on the same
analoge / digital debate , a good tool no doubt and follows the rule
" It's the player , not the gear " there's a great R & B guitarist from
Regina , Sask. Jack Semple , used an SS peavey amp for a long time
, the the rock guys in their desperation seem more suseptable to
marketing
Emperor-TK
Well-known member
I'm surprised that Digi hasn't come up with an amp modeler that needs it's virtual tubes changed every 6 months for a true authentic experience, for the low low price of $69 per tube 6 month tube license.
-Chris
-Chris
okgb
Well-known member
wow you could work for digi and of course you're right ,
and real experience would require that , and then you'd know
you would change the virtual tubes before a big session
and real experience would require that , and then you'd know
you would change the virtual tubes before a big session
bcarso
Well-known member
Did I see an eBay auction for some NOS virtual tubes?
Butterylicious
Well-known member
[quote author="Emperor-TK"]I'm surprised that Digi hasn't come up with an amp modeler that needs it's virtual tubes changed every 6 months for a true authentic experience, for the low low price of $69 per tube 6 month tube license.
-Chris[/quote]
Go ahead and take it a couple of steps further. When assembling a virtual circuit, make it so you can blow a tube. JJ can get in on the act and sell virtual tubes that red plate after 1 week or so. Of course we need used/NOS microphonic tubes. I swear that NOS software. I only booted it up on my Hickok to make sure it was good. Virtual counterfeit Mullards.
-Chris[/quote]
Go ahead and take it a couple of steps further. When assembling a virtual circuit, make it so you can blow a tube. JJ can get in on the act and sell virtual tubes that red plate after 1 week or so. Of course we need used/NOS microphonic tubes. I swear that NOS software. I only booted it up on my Hickok to make sure it was good. Virtual counterfeit Mullards.
okgb
Well-known member
that's brilliant !
A NOS plugin to modify the tube type modeling mode
quick patent the idea
A NOS plugin to modify the tube type modeling mode
quick patent the idea
7string
Member
The most valuable part of ReValver is the cabinet impulses. You can even make your own impulses and load them in.
I hooked up an old ADA MP-1 (2 x 12ax7) preamp direct. The direct distorted sound is really buzzy. Angry skeeters. Using only the programs speaker module and deleting the rest, it was shocking how much ReValver cleaned it up. It seemed like a lot more than just a lowpss. Getting some distortion outside the box definitely helps. Almost done with a Tweed Bassman / Marshall preamp to use with it. Should see some improvements.
In the tube tweaking section I tried some values that were way off and it started oscillating.
I hooked up an old ADA MP-1 (2 x 12ax7) preamp direct. The direct distorted sound is really buzzy. Angry skeeters. Using only the programs speaker module and deleting the rest, it was shocking how much ReValver cleaned it up. It seemed like a lot more than just a lowpss. Getting some distortion outside the box definitely helps. Almost done with a Tweed Bassman / Marshall preamp to use with it. Should see some improvements.
In the tube tweaking section I tried some values that were way off and it started oscillating.
