FX/Amp modelling software vs the real thing

Emp_Fab

I've been poking around with various FX & Amp modelling packages recently and I'm naturally a cynical git and got to wondering whether these apps bear any actual resemblance to the products they claim to emulate.  One major component of their appeal is the ability to 'have' loads of different pedals and amps for miniscule prices.  However, very very few folk will have the hardware equivalent to make a comparison with, so users have no real way of knowing whether the sound their getting from the modelling app is even remotely close to the real thing.

This is where the app developers have us over a barrel.  They tell us we can have the exact sound of a Bogner TwinJet through a 4 x 10 vintage cab for example, and we have no way of knowing whether the sound we're hearing is spot on or nowhere feckin near it !  All we can say is whether it is pleasing to our ears or not.  Some apps go way overboard with it - Positive Grid's Bias Amp for example - they claim to not only emulate the real thing but you can "go inside" the amp and change valve types etc, to further tweak the sound.  I find it difficult to believe that I'm listening to what a real Bognor amp sounds like to begin with, but to tell me this is what it would sound like with a 12AX7 instead of a 12AT7 ?  Really ?  Hmmmmmm.

I know these companies stick photos up of them recording the real thing in soundproofed studios etc, but isn't this just a case of the emperors new clothes - more style over substance, or do you think they are pretty good and a groundbreaking technology ?

fretmeister

In 2 parts


Part 1

I have owned Mesa Dual Recs and a few different Marshalls that the Helix models.

I had my latest Dual Rec at the same time as my Helix so I could compare. At similar volume levels it was very close - close enough to not care. I don't know if that is the case with other amps


Part 2

I can make excellent sounds with my Helix so I don't give a flying whether it is accurate or not

John_P

Most people won't have access to lots of amps but plenty will have a few to compare - I can compare the fractal models with a superbass, jcm800, jubilee and fender twin and they are close enough that I'd struggle to hear the difference in a recording.  For a while I was running into a 5153 via 4cm and I could swap the real preamp or modelled preamp - it was amazingly close - my ears aren't good enough to tell the difference in a band mix. 

The harder issue to solve is the way an amp responds and the feel while playing.  Given that people could be amplifying the modelled sounds with a variety or power section it gets tricky.  
I don't know the positive grid gear but most modelling gear is now at a level that it is incredibly close.  

digitalscream

Helix Native is the only one that's come close, in my experience, and it's way too expensive to be considered with the others. Amplitube's fairly crap, as is Guitar Rig, and BIAS...it's a massive turd as far as I'm concerned. They have loads of fanboys due to the sheer size of their marketing budget (getting lots of prominent YouTubers to promote it, thus instantly converting their fans to BIAS fans).

The question is...does accuracy really matter? There's a solid argument for using your ears instead of your eyes. It's possible to get a passable tone from Amplitube, and slightly less so for Guitar Rig. There are lots of decent VSTs out there, too - the LePou plugins are stellar for getting the tones you want (there's plenty of range there), but they shouldn't really be considered in comparison to the real thing; the "name" on the plugin really just gives you an idea of the rough features that are implemented and the levels of gain that you're going to get.

Funkfingers

digitalscream said:

The question is...does accuracy really matter?

^
This.

Even supposing that the modelling software has perfectly copped the functions of a given pedal, amplifier, loudspeaker or microphone, no account has been taken of the room in which the sound of the real amplifier will occur. 

Ultimately, most of this is academic. Either the modelled sounds inspire you to play or they do not. 

guitarfishbay

Whether it is pleasing to hear or use is the main point for me.

Some people like to have a lot of control, some don't.  An advantage with digital is if you do want the extra control it's a lot easier than learning schematics, cheaper than buying spare valves, quicker than wiring up new speakers etc.

Digital solutions are becoming more commonplace in both recording studios and on tours.  A lot of the higher end units sound close enough to a real amp through a PA system, or HiFi/monitor speakers that all the other benefits of it being digital outweigh using the real gear.

If you're playing through traditional guitar cabs there's an argument that sticking with traditional valve gear is going to be easier to get a good sound.  But if you play through other types of speakers there's probably less in it now than ever.

Sporky

Funkfingers said:

Even supposing that the modelling software has perfectly copped the functions of a given pedal, amplifier, loudspeaker or microphone, no account has been taken of the room in which the sound of the real amplifier will occur. 

I'd want all the capture to have been done in an anechoic chamber - last thing I want is to have everything sound like the back room at their office.

Room sound is now trivial to add using IRs or a real room.

guitarfishbay

Sporky said:

Funkfingers said:

Even supposing that the modelling software has perfectly copped the functions of a given pedal, amplifier, loudspeaker or microphone, no account has been taken of the room in which the sound of the real amplifier will occur. 

I'd want all the capture to have been done in an anechoic chamber - last thing I want is to have everything sound like the back room at their office.

Room sound is now trivial to add using IRs or a real room.

I don't think they capture that way anyway, I thought it was a lot of analyising the responses of the hardware (impedance curves, circuit measurements etc) and then subjectively tuned against the real thing at a later stage.  Only the Kemper really works on 'captures', and they need the room as part of that.

I understand absolutely nothing about how to code modelling software though, so I'm not sure I trust myself as a source on this either lol

fields5069

I don't see why modelled amps shouldn't get as close to real amps as CD to vinyl.

Emp_Fab

Well, I don't think the CD/Vinyl comparison is all that valid to be honest because that is just a means of reproducing a recording - the dynamics and response are fixed in the recording surely.  With an amp, the modeller has to reproduce every nuance of the way the whole amp circuit behaves in real-time.  One is just a digital representation of a signal, the other is digital emulation of a complex circuit.

stratman3142

I'm not too worried about whether ampsims sound exactly like the products they emulate, as long as they sound good. However, the names can be useful as an guide to the type of sound to expect.

Currently, I almost exclusively use S-Gear 2 and amps from Amplitube 4 Fender Collection 2. I'm probably using the latter more than S-Gear at the moment. On the odd occasion, I use Kazrog for heavier stuff. A Helix LT has recently replaced my predaboard and I like the sounds it makes.

In the case of S-Gear 2, I always seem to use the Wayfarer and I don't even know what amps that might be emulating. It just sounds good to me. But it's a bit of trial and error to find the sounds I like based on the preset names.

In my entry for the current Composition Challenge #18, I used amps from the Amplitube 4 Fender Collection 2 for all the electric guitars. I found the names of the amp emulations were useful in helping me find the type of sounds I was looking for.

The only time I recall that I've tried to emulate an exact sound is when I did an online collaboration with a couple of guys in the US. For that I used slightly tweaked versions of the Wind Cries Mary presets in the Amplitube Jimi Hendrix collection. I wouldn't claim it's anywhere near exact but, to my ears, Amplitube gave sounds that were in the right area. Others probably have more sophisticated ears than me and it will probably sound shit to them

fields5069

I know, completely different really, I realised after I posted but I'm told at work to mention everything I can think of, there are no stupid ideas etc. What a stupid idea!

With effectively limitless processing capabilities, if digital modelling can't replicate the real thing then there must be something going on inside a real amp which is unpredictable.

Roland

Funkfingers said:

digitalscream said:

The question is...does accuracy really matter?

^
This.

Individual amps vary, in fact individual valves vary, so "accuracy" is not a precise target. The challenge for any digital model is whether it responds in a useable way. Some do within tight boundaries. Some are more flexible. The same can be said about valve amps, and there used to be quite an industry in modifiying their circuits to get particular sounds or responsiveness.

There's not a yes/no answer to whether digital modelling works.

Emp_Fab said:

... do you think they are pretty good and a groundbreaking technology ?

Our current set list requires me to use three different amps, with a range of overdrive and EQ settings. It's taken many years to get to this point, but I now carry in two hands what would previously have required a roadie and a van. 

Emp_Fab

digitalscream said:

Helix Native is the only one that's come close, in my experience, and it's way too expensive to be considered with the others. Amplitube's fairly crap, as is Guitar Rig, and BIAS...it's a massive turd as far as I'm concerned. They have loads of fanboys due to the sheer size of their marketing budget (getting lots of prominent YouTubers to promote it, thus instantly converting their fans to BIAS fans).

The question is...does accuracy really matter? There's a solid argument for using your ears instead of your eyes. It's possible to get a passable tone from Amplitube, and slightly less so for Guitar Rig. There are lots of decent VSTs out there, too - the LePou plugins are stellar for getting the tones you want (there's plenty of range there), but they shouldn't really be considered in comparison to the real thing; the "name" on the plugin really just gives you an idea of the rough features that are implemented and the levels of gain that you're going to get.

I get what you say.  At the end of the day, it's not about accuracy really - but that is the main selling point of the product.  Without that claim, I'm sure sales would be nowhere near as high.

Expensive hardware (Helix) aside, what has been your least disliked software emulator ?  Your post above lists three as fairly crap, fairly crap and 'a massive turd' :-)  Not exactly endorsements for any of them ! :-)  Have you tried any others ?

guitarfishbay

Currently my favourite is S Gear 2, or TSE X50 for a specific type of high gain sound (6505).

I'm not really getting on with Helix Native so well for guitars currently, I feel like I'm messing with the advanced parameters a lot trying to get the feel/dynamic response I like.  I've got it set up correctly (input levels), it just feels tame to play compared with my Axe FX 2, and it's a lot more expensive to buy than the software I've found that I like.  There are decent sounds in it though, and it's good for bass.

Drew_TNBD

Emp_Fab said:

One is just a digital representation of a signal, the other is digital emulation of a complex circuit.

Which in turn is just digital representations of one or many signals.

digitalscream

Emp_Fab said:

digitalscream said:

Helix Native is the only one that's come close, in my experience, and it's way too expensive to be considered with the others. Amplitube's fairly crap, as is Guitar Rig, and BIAS...it's a massive turd as far as I'm concerned. They have loads of fanboys due to the sheer size of their marketing budget (getting lots of prominent YouTubers to promote it, thus instantly converting their fans to BIAS fans).

The question is...does accuracy really matter? There's a solid argument for using your ears instead of your eyes. It's possible to get a passable tone from Amplitube, and slightly less so for Guitar Rig. There are lots of decent VSTs out there, too - the LePou plugins are stellar for getting the tones you want (there's plenty of range there), but they shouldn't really be considered in comparison to the real thing; the "name" on the plugin really just gives you an idea of the rough features that are implemented and the levels of gain that you're going to get.

I get what you say.  At the end of the day, it's not about accuracy really - but that is the main selling point of the product.  Without that claim, I'm sure sales would be nowhere near as high.

Expensive hardware (Helix) aside, what has been your least disliked software emulator ?  Your post above lists three as fairly crap, fairly crap and 'a massive turd' :-)  Not exactly endorsements for any of them ! :-)  Have you tried any others ?

The magic isn't really in the amp sims - there's good and bad, but most can be rescued with a great cab sim (same as for most hardware), which means impulses.

Of all of them, I quite like S-Gear (again, not cheap), but the ones I reach for every time I can't be arsed setting my rig up are the LePou plugins. Free, simple, and do the job.

Of course, this is apples-to-oranges. 90% of guitarists will compare an amp sim to an amp in the room; there's a huge part of the signal chain missing there. The amp sim, by definition, must include a microphone (all of which sound different). Listening in the room, your ears perceive it totally differently. They're also never in a position that can be emulated accurately by software - they're usually at least a couple of feet above the central axis of the speaker, to the side and a few feet away. That means that, by listening to an amp in the room, you're hearing far more of the room's influence on the sound than you think.

Give that same 90% of guitarists an amp in an isolated room with a mic in front of the cab and get them to listen to it from the control room, and they'll tell you it's not a real amp because "it doesn't sound or feel the same".

steer

My only real experiment was to take a Zoom MS100bt with the default TS9 Tubescreamer, and run it back to back with a real TS9. I could not tell the difference.

My Zoom is relatively old tech now, so I guess the digital stuff will carry on getting better every year. 

I've said this before - but when you are playing in a band, with drums, bass and keys etc, you are part of the mix and I would bet that no one in the crowd will be able to tell whether you are playing though a boutique amp or a good quality digital emulation. 

Emp_Fab

fields5069 said:

I know, completely different really, I realised after I posted but I'm told at work to mention everything I can think of, there are no stupid ideas etc. What a stupid idea!

With effectively limitless processing capabilities, if digital modelling can't replicate the real thing then there must be something going on inside a real amp which is unpredictable.

This is the key point for me.  I don't think it's a case of things going on that are unpredictable, but the sheer complexity of an analogue circuit that I have trouble believing can be accurately emulated.  It's a computer program at the end of the day...  page after page of code that is aiming to alter an analogue input signal in an identical manner to that of the real amp.  That sounds difficult enough, but when you consider that it has to do this for every conceivable combination of settings on the real amp - AND for every conceivable type and quality of input signal...  I've done a (very) little bit of coding but I just can't get my head around how something logical like a computer program can deal with all the variables involved.  Variables that vary in the way they vary depending on the circumstances !  I'm not saying that 100% accurate emulation of the behaviour and qualities of a real amp is impossible, just that I imagine it might take a room full of coders months to achieve and then it would be gargantuan.  To do that for multiple amps... with multiple cabs.... and multiple effects... all of which interact in many different ways seems like an impossible task.

I obviously don't really know what the processes are in emulation software, but I'm left thinking "they can't do that....  it's insane".  That would be like creating artificial intelligence that could pass as human to me !

steer

Valves are one thing, but anything that was originally solid sate is basically silicon to start with anyway. 

Drew_TNBD

Emp_Fab said:

fields5069 said:

I know, completely different really, I realised after I posted but I'm told at work to mention everything I can think of, there are no stupid ideas etc. What a stupid idea!

With effectively limitless processing capabilities, if digital modelling can't replicate the real thing then there must be something going on inside a real amp which is unpredictable.

This is the key point for me.  I don't think it's a case of things going on that are unpredictable, but the sheer complexity of an analogue circuit that I have trouble believing can be accurately emulated.  It's a computer program at the end of the day...  page after page of code that is aiming to alter an analogue input signal in an identical manner to that of the real amp.  That sounds difficult enough, but when you consider that it has to do this for every conceivable combination of settings on the real amp - AND for every conceivable type and quality of input signal...  I've done a (very) little bit of coding but I just can't get my head around how something logical like a computer program can deal with all the variables involved.  Variables that vary in the way they vary depending on the circumstances !  I'm not saying that 100% accurate emulation of the behaviour and qualities of a real amp is impossible, just that I imagine it might take a room full of coders months to achieve and then it would be gargantuan.  To do that for multiple amps... with multiple cabs.... and multiple effects... all of which interact in many different ways seems like an impossible task.

I obviously don't really know what the processes are in emulation software, but I'm left thinking "they can't do that....  it's insane".  That would be like creating artificial intelligence that could pass as human to me !

https://en.wikipedia.org/wiki/SPICE

There ya go.