Modeler Coding.

[El Dunco]

I was doing a lot of thinking today about how modelling could be great if the developers tried to create their own ideal amp instead of trying to imitate an existing amp the best... An amp sounds the way it does because of the way it's built and there's no simulating that.

However... the technology could have you create a sound that you wouldn't have thought possible and I realized... The drummer in my band and close friend is a programmer and our bassist (who works with him) is a graphic designer! Together we could be a dream team!

Ryu (drummer) and I are going to tinker around with some of my equipment and some software and start developing amp models.

We're not just stealing open source and changing it around, we're starting from scratch here.

I'd like to hear from you guys what you just wish they had in amp models that just isn't a thing yet and we'll see what we come up with and when we've got our first prototype, I'll share it with the forum and see what you guys think!

Do comment.

 

[dominus]

Re: Modeler Coding.

Ideally -

EQ - Distortion - EQ - Tube Warmth - Speaker Cabinet

 

[dominus]

Re: Modeler Coding.

Also, these guys do some emulations, but they're also doing original designs:

http://lepouplugins.blogspot.com/

http://www.igniteamps.com/audio-plug-ins

http://www.tseaudio.com/software.html

They all sound incredible and they're all donationware.

 

[El Dunco]

Re: Modeler Coding.

Yeah that was pretty much idea. The ability to shape the tone before and after the distortion and the ability to load impulses into the client itself and adjust the reactance in the same way you'd match presence and resonance to real cabinet.

The best modeler of a Marshall will sound like an excellent recording of a Marshall in a studio, but we're going for something different. We want to use the technology to make something new that takes advantage of the technology.

There's an amp called Voxengo that has a neat idea, but it's execution is terrible and borders on being unusable.

Also how's this for an idea? Interactive impulse responses.

If impulse responses are like a snapshot, I'm going for a .gif. Some way for the software to have the impulses react to the input sound the way a speaker would. It wouldn't be too hard to do, it kind of works on the same principle as lights that "dance" to music when you plug them into the speaker outs of a stereo. That would deal with the "static" feel a lot of people point out about impulses.

 

[dominus]

Re: Modeler Coding.

Yeah that was pretty much idea. The ability to shape the tone before and after the distortion and the ability to load impulses into the client itself and adjust the reactance in the same way you'd match presence and resonance to real cabinet.

Alu said that he might be working on something like that to go along with the TPA-1 power amp emulator.

The best modeler of a Marshall will sound like an excellent recording of a Marshall in a studio, but we're going for something different. We want to use the technology to make something new that takes advantage of the technology.

There's an amp called Voxengo that has a neat idea, but it's execution is terrible and borders on being unusable.

Boogex? Yeah, Alexei is a master coder, but he's not a guitarist.

Also how's this for an idea? Interactive impulse responses.

If impulse responses are like a snapshot, I'm going for a .gif. Some way for the software to have the impulses react to the input sound the way a speaker would. It wouldn't be too hard to do, it kind of works on the same principle as lights that "dance" to music when you plug them into the speaker outs of a stereo. That would deal with the "static" feel a lot of people point out about impulses.

Have you checked out Revalver's RIR module? It lets you load impulses, adjust the EQ (Lows and Highs knobs), the distortion of the speaker and mic (Dist and Crunch knobs) and the resonance and presence (Bottom and Brilliance knobs):

Even though I own Recabinet, I always find myself going back to using RIR to load and tweak impulses. It just feels more natural to me.

 

[El Dunco]

Re: Modeler Coding.

Close, but a step further. Those are all ways of colouring the impulse. I'm talking about a way that has the impulse portion actually subtly respond to what it's being fed. Even if we start with something as simple as having slight volume fluctuations in reaction to the incoming waveshape. It would feel a bit like a speaker moving back and forth.

 

[beaubrummels]

Re: Modeler Coding.

I want a modeler that allows me to change sets of tubes and tweak them, run them hot or normal. If it's virtual, why can't I do things you can't now, like naively try every tetrode/pentode there is in a Marshall Super Lead circuit (disregarding plate voltages and heater filament draw, and all the incompatibilities ), or change the solid state rectifier to a GZ34 to hear the affect, or put a Hiwatt tone stack in a Marshall head to see what difference it makes (if any), or swap all the caps in the circuit from type to another.

 

[El Dunco]

Re: Modeler Coding.

I want a modeler that allows me to change sets of tubes and tweak them, run them hot or normal. If it's virtual, why can't I do things you can't now, like naively try every tetrode/pentode there is in a Marshall Super Lead circuit (disregarding plate voltages and heater filament draw, and all the incompatibilities ), or change the solid state rectifier to a GZ34 to hear the affect, or put a Hiwatt tone stack in a Marshall head to see what difference it makes (if any), or swap all the caps in the circuit from type to another.

That sounds like a REALLY long term project... but with this software that Ryu has if I can capture enough elements of different components and add them to the 'collection' it's definitely feasible.

Have you tried Revalver?

 

[dominus]

Re: Modeler Coding.

I want a modeler that allows me to change sets of tubes and tweak them, run them hot or normal. If it's virtual, why can't I do things you can't now, like naively try every tetrode/pentode there is in a Marshall Super Lead circuit (disregarding plate voltages and heater filament draw, and all the incompatibilities ), or change the solid state rectifier to a GZ34 to hear the affect, or put a Hiwatt tone stack in a Marshall head to see what difference it makes (if any), or swap all the caps in the circuit from type to another.

Revalver lets you do all that.

 

[dominus]

Re: Modeler Coding.

Yeah that was pretty much idea. The ability to shape the tone before and after the distortion and the ability to load impulses into the client itself and adjust the reactance in the same way you'd match presence and resonance to real cabinet.

The distortion has to react with the guitar correctly though. I think that's Boogex's problem. I messed around with some non-guitar specific distortion plugins yesterday and they just didn't react the same way a pedal does. (I'm talking cheap distortion pedals that sounded 1000 times better.) I've had luck with just about every analog distortion pedal in my chain. (Digital ones seem to be hit or miss. As much as I like my plugins, digital pedals seem to be adding a tone that I can't quite get around. Guitar pedal plugins do react the right way. It must have something more to do with *how* it clips, rather than just distorting.) I'm even tempted to grab some Danelectro and Behringer pedals *just* to try out.

This is one of the videos that gave me the idea:



but I go one step further.. I've got the EQ, power amp, and cabinet all being done virtually on the computer. Yes, I do have the Badass Distortion, and yes, it does sound that good being used this way. I think I paid $50 for it. Not bad for a preamp.

 

[Hank-]

Re: Modeler Coding.

Scuffham S-Gear ver.2 that I tried once I really liked. It had a really good sounding delay plugin too. It's not very expensive either. Never tried revalver, the possibilities look really cool from the screenshot above.

 

[JeffB]

Re: Modeler Coding.

Roland came out with its own modelling combo amp last year that was designed to do your own thing with andnot mimic old amps. It tanked apparently...took sweetwater a long time to get them,.and then they were on blowout 6 months later. Weird. Never saw one myself. Might wanna check em out.

 

[uOpt]

Re: Modeler Coding.

We're not just stealing open source and changing it around, we're starting from scratch here.

"Stealing from open source"? The whole point of open source (in the narrow sense of the word, not just "can read the source") is that you are supposed to re-use and build on it.

Now, if you want to take your product private and not make your modifications to an open source project available as open source in turn - that still works. Almost all open source software comes under one of two license families: 1: GPLed, so you are not allowed to do that 2: BSD/MIT license where you are explicitly allowed to do that.

I wouldn't believe you don't at least start on top of FFTW or similar.

 

[uOpt]

Re: Modeler Coding.

As for the original question.

Forget amps.

What we are missing is a highly sophisticated speaker simulator, just for starters one that takes into account what the output impedance of the amp is (because a lower output impedance puts a "brake" to the speaker's movements after the impulse ends). Unfortunately that quickly goes into the zone of microphone simulation and now you have a room in between.

 

[ratherdashing]

Re: Modeler Coding.

Go way outside the box. Make me a modeller that makes my guitar sound like an orchestra, or a synth, or a choir, or something I have never heard before. Give me something I can't even begin to get from current modelling gear.

 

[GuitarStv]

Re: Modeler Coding.

I did my final engineering project building a reprogrammable board that did sound effects. Chorus, reverb, flanging, phasing, delay, distortion . . . it's all pretty straightforward once you figure out how you need to modify the sound wave to achieve the desired effect. Do you want to do the modelling real-time on the fly, or just take a clean guitar signal, process it, and then output the new result?

 

[dominus]

Re: Modeler Coding.

Roland came out with its own modelling combo amp last year that was designed to do your own thing with andnot mimic old amps. It tanked apparently...took sweetwater a long time to get them,.and then they were on blowout 6 months later. Weird. Never saw one myself. Might wanna check em out.

Are you talking about the Cube?

 

[JeffB]

Re: Modeler Coding.

Are you talking about the Cube?

Heh. Nope. Its a GA-112 or something like that.

 

[uOpt]

Re: Modeler Coding.

I was just about to ask this, so why not here.

This is about bass, not guitar.

I don't want 238 gimmick effects. I want:

• A programmable and midi-recallable 4-band fully(!) parameter EQ or 15 band graphic EQ
• Plus(!) low and high shelf EQ with 2 controls (frequency and amplitude)
• Plus HPF (pop filter) and LPF (better than nothing speaker simulation)
• A programmable and midi-recallable compressor, with knee control and separate attack and release time control
• Built-in frequency crossover (additional outputs), and you know what, why not make it multiband compression if you have the crossover anyway?
• 4 or more effect loops so that I can stuff my reverb and chorus and whatever else in
• And the order of effects including the loops is freely programmable

A TigerSharc CPU would woop that with ease (probably with < 10% CPU load) and it would be an incredibly useful tool.

 

[El Dunco]

Re: Modeler Coding.

"Stealing from open source"? The whole point of open source (in the narrow sense of the word, not just "can read the source") is that you are supposed to re-use and build on it.

Now, if you want to take your product private and not make your modifications to an open source project available as open source in turn - that still works. Almost all open source software comes under one of two license families: 1: GPLed, so you are not allowed to do that 2: BSD/MIT license where you are explicitly allowed to do that.

I wouldn't believe you don't at least start on top of FFTW or similar.

No need to go off at me. I'm not the expert at coding here... Ryu is.

All I meant is that we're not going to simply modify existing a modelling software program, we're building ours from the ground up. How is that a bad thing?