Think you can tell real amps from models?

ezietsman

Just saw this today.

http://www.seymourduncan.com/tonefiend/recording/can-you-tell-amps-from-amp-models/

ezietsman

I listened to the clips last night, I'm having difficulty deciding which clip corresponds to which amp, nevermind which clips are models.

The challenge is carefully designed to be difficult, he didn't really use the 'classic' tones one would expect from each of these amps, the sounds on each clip can be produced by any of those amps with some careful knob-twisting.

Jack-Flash-Jr

By the time it comes through my laptop speakers it's been processed so many times it might as well be modelled. I think live is the real test...

MIKA-the-better-one

I think I could easly tell amps apart..... the models and amps are so close and sound pretty much the same.... non the less an amp is cooler

Gearhead

Listening to someone play an amp or model is harder to distinguish than playing amp or model yourself, so any recording test is flawed.

MikeM

As usual, I believe the magic of a tube amp it in its dynamism..

needleshy

I was listening to heory of a dead man the other day and I was thinking "damn those are cool sounding amps they are using", so i googled it and found they are using didgidesign eleven racks, i thought it was tube amps :-[

Keira-WitherKay

i used to be a huge "modeller " fan ..... as i travelled a lot and gigged and a small modelling preamp is easier to carry ..... BUT in last 9 months i been using tube amps on stage for my gigs ,

and at moment if i say play on an amp modeller thru the PA for say an hour i get so used to the the sound of the modeller say doing my favourite amp tone , that i start thinking "wow" this sounds almost like the tube amp why do i bother dragging an amp around( cos i'm intrinsically lazy and won't carry extra gear just for kicks) .

when doing a direct A/B ......there's something in the tube amp sound that i like and oddly enough i play with the amp mostly clean ..so it's NOT just a distortion/overdrive tone thing but decidedly also a clean sound tone thing where the tube amp offers some mojo , i'm still to experience in modelling amps

but on the other hand i could use a modeller at any gig i do and i'm sure no one but me would even notice or care , ok maybe a few ?

Tokai-SA

Here's a good example related to this thread.

Not even the guitarist who's amp is being profiled could tell which was which.

shadow1

SCARY clip that! so there might just be hope that in the future we wont need to lug around heavy massive amps to get the tone we desire ?

best modelling I have heard

ezietsman

So there's a good bunch of us that can't tell the clips apart. LOL.

I also think that today, given a model of a, say, Deluxe Reverb and a real one, we'd be able to discern the real one when we play it (lets assume they mic it and put it behind a wall so you cannot see what you are playing on). And I think most people will not be able to tell. In a few years we'll have sooper dooper mark VIII intel Core I50s with so much ram you store wikipedia in memory. when that time comes, there's going to be nothing stopping a model behaving EXACTLY like a real amp. When computers can solve, in real time (i.e. 44 000 times per second), the full set of differential equations representing the circuit in your favourite t00b amp, the difference becomes academic.

Still, its fun carrying around a 25kg 2x12 cab and a tube amp head filled with copper coils and heavy transformer cores every time you want to jam. And it looks cool ???

Edit:

Another thing that I like about valve amps, besides their cool looks and light weight (LOL!) , seriously, is their simplicity. Any concievable modelling device, however good sounding will be more complex than most classic valve amps. Then the question is reliability and ease of mending (amps break, all of them)

Wizard

I just like tube amps.

BMU

Gearhead wrote: Listening to someone play an amp or model is harder to distinguish than playing amp or model yourself, so any recording test is flawed.

That's probably true. Which means the test isn't flawed from a listener perspective. I.e. only the guitarist cares, not his audience.

Manfred-Klose

i spotted 2 samples right away that was plugins, but could not really tell the rest apart.

Its funny if you think about, amps and plugins keep evolving, but all they try to do is t replicate the sound of classic amps.
in many years from now, when stuff has gone digital, they'd still be trying to replicate the classic sound.

i am happy my Guitar(tuned properly) >>>> decent guitar cable >>>> good sounding valve amp.

"Technology killed creativity"

henrebotha

Wizard wrote: I just like tube amps.

I think any argument except this one is crap.

ezietsman wrote:When computers can solve, in real time (i.e. 44 000 times per second), the full set of differential equations representing the circuit in your favourite t00b amp, the difference becomes academic.

Do you know that models are not doing this already? Not a rhetorical question, I actually want to know. ?

Gearhead

I suppose they do, but in general modelling means simplifying. You can neglect some of the more subtle factors of your equations or simply not be aware of them, which then makes your model differ from reality. For instance, with the newtonian force model you cannot explain why bath water swirls around the drain (clockwise on the northern hemisphere and the other way over here) but if you take the Coriolis acceleration into account it is all too clear.
One way to avoid having to figure out all the small things is to not model, but rather work with statistical data of how sound gets changed. Profiling amps work that way as far as I understand the thing.

ezietsman

Gearhead wrote: I suppose they do, but in general modelling means simplifying. You can neglect some of the more subtle factors of your equations or simply not be aware of them, which then makes your model differ from reality. For instance, with the newtonian force model you cannot explain why bath water swirls around the drain (clockwise on the northern hemisphere and the other way over here) but if you take the Coriolis acceleration into account it is all too clear.
One way to avoid having to figure out all the small things is to not model, but rather work with statistical data of how sound gets changed. Profiling amps work that way as far as I understand the thing.

I think you're right about the statistical models. I think they have a kind of generalised model inside the profiling amp (seeing that it doesn't know what kind of amp it is profiling) then taking the input, which it gets from your guitar and the output, which it get's via the mic or line-out, it works out, empirically, some numbers which represents the transformation from the one to the other. This is a well established way of making a model of something if you need numbers, which for making noise is fine. The big hassle with such systems are to find a mathematical description that:

1. allows enough degrees of freedom to match the output to the input
2. to do 1. whilst staying fairly well-behaved in regions not covered by the input.

If this sounds weird think of it like this:

you are profiling your own amp. You set it for a very overdriven tone but roll your guitar back so that it is clean. Now you play some quiet chords and such and the profiler matches that input up to the output. Fine. Now you switch to the profiler and turn your guitar up to 100%. What happens? In this case you would like your guitar sound to not be TOO zany but the model doesn't have much knowledge of what should happen, but it should make a more distorted tone i.e. it should behave like you'd expect hence ''well-behaved''.

Doing a good profile of an amp shoudn't involve any playing really. The profiler should be able to inject signals into your amp all by itself and build up a complete empirical model of all (most) expected inputs. I think it does this anyway (when its playing those crazy space sounds). However, given the number of different amp designs and such, I'd be highly surprised if it can model any one amp well over its entire range of sounds, even with generous input info. It will excel at covering a smaller set of range of sounds per amp, because it is easier to model a small domain than a complete one.