Swanky Amp (release 1.4.0)

[jbraner]

Thumbs up for 0.10.2!
Sounds good here. I do hear the sag doing a little something - I'm just using a little bit.
Another thing I noticed - the CPU for this amp is low. It's half that of my Neural DSP stuff!

I hope you'll be getting the champagne out when this goes to ver 1.0, and I still think you should be charging $50 - $100 for this

Also, I hope you'll have a mailing list, or else use this thread to keep us informed re: what's going on with this project.

[garrinm]

just testing this plugin, i really like the clean tone

Thanks for dropping a line, I'm glad your liking the plugin. From the video it looks like you're clean tone was just using a bit of power amp drive and some sag. The sag can really soften things up by adding a transparent compression. At high drives though that compression becomes ... not so transparent xD.

I noticed you didn't gain stage your input. You might find you get a better range of drive if you set your input level to match your pickup / audio card amplification. The tooltip on the input level knob tells you what to do there, but in a nutshell you want to increase the input level until the input level meter is peaking right over the H notch (S if you're playing a single coil pickup).

Garrin

[garrinm]

Thumbs up for 0.10.2!
Sounds good here. I do hear the sag doing a little something - I'm just using a little bit.
Another thing I noticed - the CPU for this amp is low. It's half that of my Neural DSP stuff!

I hope you'll be getting the champagne out when this goes to ver 1.0, and I still think you should be charging $50 - $100 for this

Also, I hope you'll have a mailing list, or else use this thread to keep us informed re: what's going on with this project.

Glad to hear this patch is working out! Thanks again for the kind words.

I've been keeping a close eye on the CPU and I've got a few optimizations in there. But really I think this comes from 2 big factors: 1) FAUST creates very efficient loop code and 2) there's no expensive impulse responses in there. This afforded the room to really pack a bunch of non-linear processing in the tube models and not skimp on the gain stages.

I've started work on a pro version and I'm excited to share that as I progress. When I release 1.0 I'll put out the blog post and probably create a mailing list, good idea! I just need to put some time aside to do these things but at the moment I'm really excited about the work for the pro version.

I'm not sure what the overall communication strategy will be, but for the near future I think this thread is the primary home of everything Swanky Amp.

Garrin

[garrinm]

"…quietly played sections are bright and full, but when the sound gets very loud and distorted, the cab sort of 'closes in on itself' (the upper shelf moves back a bit), making the distortion less fizzy and a bit softer" is as good a description of O.T. behavior as I've seen anywhere.

If it's not "physical model" enough for inclusion in the amp section, maybe it could be labeled "Output Dynamics" and be separated from the cab sim (cab having its own seperate bypass switch, the idea being that you can use the Dynamics with any cab sim or IR)? Just spitballing ideas here. Including an IR loader would also do the trick, but that sounds like a lot more work and it's not as flexible for post amp sim routing.

Hi guitarzan, thanks for those ideas. The cab model right now is based on the frequency response of a bunch of impulse responses I found online. I also own the Celestion pack, and that one has responses measured at "lo gain" and "hi gain". So for the dynamic cab what I did is I measured how my model for what a 4x12 cab changes at high gain. The biggest effect is the shelf moving down, but also some boosting in the lows as well as some absorptions that seem to shift around a bit. I think these effects would differ in magnitude and where they occur in the frequency spectrum for different cabs. So in the end I don't think I have a general model of how any cab response should change at high gain, but rather just a decent model for how a 4x12 cab response changes at high gain.

That said, it's true I could separate that effect from the cab model, so that if you plug in your own cab IR you still get those dynamics. But they would be wrong unless your cab IR is a 4x12 cab similar to the ones I used to derived my model. Probably it would sound just fine though, so it's something I'm considering.

As for the output transformer, I'll keep my eyes open for resources that might help me do something there. But without some measurements or a good mathematical description of those non-linear behaviors there's not too much I can do at the moment on that side.

Thanks,
Garrin

[jbraner]

I've been keeping a close eye on the CPU and I've got a few optimizations in there. But really I think this comes from 2 big factors: 1) FAUST creates very efficient loop code and 2) there's no expensive impulse responses in there. This afforded the room to really pack a bunch of non-linear processing in the tube models and not skimp on the gain stages.

FYI - I was comparing it to another amp sim - with the cab disabled (like it's disabled in Swanky Amp) and you still come out at <= half the CPU

That said, it's true I could separate that effect from the cab model, so that if you plug in your own cab IR you still get those dynamics. But they would be wrong unless your cab IR is a 4x12 cab similar to the ones I used to derived my model. Probably it would sound just fine though, so it's something I'm considering.

That would still really be worth looking into. If it's good - maybe we could see how to derive curves from other cabs (or cab sims)

[guitarzan]

…So for the dynamic cab what I did is I measured how my model for what a 4x12 cab changes at high gain. The biggest effect is the shelf moving down, but also some boosting in the lows as well as some absorptions that seem to shift around a bit. I think these effects would differ in magnitude and where they occur in the frequency spectrum for different cabs. So in the end I don't think I have a general model of how any cab response should change at high gain, but rather just a decent model for how a 4x12 cab response changes at high gain.

That said, it's true I could separate that effect from the cab model, so that if you plug in your own cab IR you still get those dynamics. But they would be wrong unless your cab IR is a 4x12 cab similar to the ones I used to derived my model. Probably it would sound just fine though, so it's something I'm considering…

Maybe too much more work for this version, but like jbraner said, if you did the same type of IR comparison with typical 1X12, 2X12, 4X10, etc you could probably come up with a tweakable cab response sim that would work great for any IR and far surpass the "close enough for rock 'n' roll" standard (what you have now probably already would), and it really does cover pretty much the same ground as an O.T. as far as audible tone alteration (warm/thicken and smooth out the jagged highs) — in fact, if you used IRs that were made using tube power amps as I believe the Celestion ones are, you would already be including audible O.T. artifacts in your comparison and resulting model.

The cab model right now is based on the frequency response of a bunch of impulse responses I found online.

While I really dig the new dynamics, the new speaker sim itself has a phasey comb filtery thing going on that I really don't like*, similar to what I often hear in IRs made using sm57 mics. So it's a personal taste thing I suppose, which is why I think separating the output dynamics processing from the actual speaker sim would make for a much more versatile amp model.

I'll be checking out the latest update ASAP, hopefully sometime today or tonight. There is always some sag going on regardless of Sag setting, right? Just more transparent the lower the setting? I noticed in that posted video that I clearly hear sag regardless of Sag setting even though sometimes I don't always notice it when I'm playing at similar settings. I guess I am getting more used to it and it's one of those things you notice more when it's absent — other sims seem so stiff, to the point if being unplayable (to me, anyway).

* — if the cab dynamics processing is splitting the signal maybe this is introducing the comb filter thing unintentionally? If you weren't intentionally trying to mimic that mic phase sound, then maybe look into adjusting phase alignment.

If the signal does need to be split for parallel processing then the output dynamics section probably would need an onboard cab sim or IR loader and a manual phase alignment control (Boogex has one I think). Parallel processing would be tricky with an external IR loader or cab sim.

[guitarzan]

I tried version 0.10.2 late last night and it's hard to comment not being able to A/B versions, but initially I thought the tone changed some — like more strident mids or something, less saturated at the same settings in the previous release, but maybe it all comes down to just the sag set point changing — so maybe sag is now less from 5 down and more from 5 up? So it's kind of a wash — I like the extended range for sag, but now seems too strident (for me personally) set below noon (a bit brittle for my taste, transistor-like). The shape of the sag envelope also seems a little different (more tame?) than the older versions where you could set set sag to be more apparent like in this latest version?

So it's fine I guess if others are happy with the less apparent sag at lower settings — there is still plenty up top for me. Like I said, hard to tell if my impressions are based in reality or not without A/B confirmation.

EDIT: After re-reading all that, one of the things I was trying to get across is that there is some point in the Sag setting where I still clearly hear sag in a recording, but no longer experience it as "feel" while playing. That point seems to have moved to a higher setting on the dial in this update.

Also, I'm not sure how useful the highest sag settings really are but it was kind of cool to be able to hear the envelope clearly again. Running one Swanky Amp with high sag in parallel with another set tighter might sound really fat — probably should give that a try (keeping phase in mind)!

[guitarzan]

So to summarize, dancing about architecture is a color of sound that feels just like it tastes.

[garrinm]

I'll take the silence as a sign the last patch has been working pretty well xD. The silence from my side has been because I've had my head down working on the starts of the pro version and revamping the website in preparation for the release of 1.0.

I've finally finished the blog post I've mentioned a few times. It turned into a blog series, with the last 2 parts being the most relevant probably to this audience. You can find it here:

http://www.resonantdsp.com/blog/tube-em ... ion-1.html

I'm hoping to release 1.0 tomorrow, pending a few things. I'll really just be 0.10.2 with an additional tone stack! And the power amp drive will default to a lower value when choosing the init preset. But otherwise it should perform the exact same and all your presets should be unchanged.

[prodigal_sounds]

Very enjoyable blog series. I agree that it seems time for a v.1.0.

[acousticglue]

I keep getting oddly sticking stage dial

[guitarzan]

Yeah, great blog, especially pt.4. Plus the home page ABOUT section… Wow, Garrin the PhD in particle physics — I did not see that coming! I think we have the right guy for modeling nonlinear dynamics!!

I keep getting oddly sticking stage dial

I noticed that the STAGING adjustment was different, but I figured it's just supposed to index between settings now rather than smoothly rotating?

[jbraner]

I'm looking forward to reading the blog. I'm just in the final stages of a PC upgrade at the moment - so a little busy

[jbraner]

Well, that was a good read!
Also, I signed up for the "Pro" newsletter - as I'd like to support the project.

I'm almost ready to start a new project - so 1.0 will be out by then, and not likely to change - so i'll be able to use Swanky Amp for some or all of my guitar parts 9there are always guitar parts).

Keep us updated!

[garrinm]

Alright, good news: I've got Dave Clark's Marshall-like tone stack integrated in the amp and it sounds... like a Marshall tone stack. It's a simple switch but it goes a long way to add variety to the tones possible. I just have to re-work the presets and make the installers, then I'll drop the update.

I'm glad people are enjoying the blog series. I'll be adding more to it in the coming weeks. I'll probably devote a blog post to sag; I'll also be putting some updates about the pro version in there; and I'd like to do something about guitar tones in general and bringing it back to the basics of what is distortion and whatnot.

I keep getting oddly sticking stage dial

It's like guitarzan said: in the 0.10.0 update many things changed under the hood and I had to change the gain staging to be a selection instead of allowing in-between values. I didn't find it that useful, but I might bring that feature back in. I think it'd be interesting to blend the tone stacks this way.

I'm almost ready to start a new project - so 1.0 will be out by then, and not likely to change - so i'll be able to use Swanky Amp for some or all of my guitar parts 9there are always guitar parts).

Keep us updated!

It should be the same amp you're used to, but with the ability to switch to a Marshall-like tone stack. For when you need those mids. And you need lots of them. Can't wait to hear what you do with it.

Thanks,
Garrin