Asher Guitars WD Music Products Amplified Parts Mod Kits DIY Nordstarnd Pickups Warmoth.com
Asher Guitars WD Music Products Amplified Parts Mod Kits DIY Nordstarnd Pickups Warmoth.com
Asher Guitars WD Music Products Amplified Parts Mod Kits DIY Nordstarnd Pickups Warmoth.com

What is a "Modeling" Amp?

Discussion in 'Modeling Amps, Plugins and Apps' started by JayFreddy, Mar 1, 2013.

  1. Gnobuddy

    Gnobuddy Friend of Leo's

    Sep 15, 2010
    British Columbia
    Thanks for that link! I wasn't aware of Google Scholar, which may be why I had so little luck finding anything technical enough to be useful.

    I just got through a first read of that survey paper. It was rather disturbing to find out just how crude most of the attempts to model nonlinear transfer functions are, at least as described in this paper. No wonder most of the modelling amps I've heard sound so harsh.

    One of the techniques mentioned in the paper - using a piecewise linear approximation to a nonlinear transfer function - has been used in analog guitar electronics decades ago. An opamp with a network of resistors and diodes in the feedback network can be used to generate the piecewise linear
    curve. It actually works better than a purely digital model would, because the diodes slightly smoothen the transitions from one straight line segment to the next, reducing generation of high order harmonics.

    I remember a circuit for a guitar preamp published in Elektor magazine some time in the early 1980's that used this technique to generate a soft-clipping distortion stage.

    Back to DSP and digital modelling, I suspect the techniques used in todays state-of-the-art modelling devices like the Kemper Profiling Amplifier and AxeFX II are considerably more complex than the methods discussed in that 2009 review paper. I know that physical modelling has been quite successful in synthesising piano sounds lately ( http://www.pianoteq.com/ ), and I wonder if vacuum tube amplifier stages are now being physically modelled in the same way Pianoteq models piano wire, piano hammers, and piano soundboards. That's likely to work much better than crude lookup tables and interpolation formulae.
    I could certainly see a code module that simulates, say, a single 12AX7 common cathode gain stage. Plug in resistor and cap values, chain a few of them together with FIR filters for the tone controls, and you'd be on your way to building a virtual amp.

    -Gnobuddy
     

  2. JayFreddy

    JayFreddy Poster Extraordinaire Silver Supporter

    Age:
    52
    Nov 6, 2006
    Dallas TX USA
    I believe that's what Yamaha claimed to do with their DG series amplifiers and their proprietary ECM (Electronic Circuit Modeling) technology.

    The Drive 1 setting on my DG80 112a is still one of the most dynamic, touch sensitive overdrive sounds in my arsenal, and that includes a bunch of real tube amps and boutique overdrives.

    One thing I've noticed since I started this thread, is fewer amp manufacturers are using the term "modeling" to describe their digital offerings... Some still do, but the trend seems to be away from that terminology.

    A Mustang II uses, "an all new DSP platform"...

    Works for me. ;)
     

  3. Gnobuddy

    Gnobuddy Friend of Leo's

    Sep 15, 2010
    British Columbia
    I agree on all counts. I haven't heard any modelling products released lately that sound as bad as my horrible Line 6 Pod 2.0 or Spider III sounded ten years ago.

    Then again, in 1999 my Pentium II computer had a single-core CPU running at 200 MHz and a 2-gigabyte hard drive. Today you can buy a four-core Android mini-computer clocked at over 1 GHz and about the size of a pack of chewing gum - and you can buy it for under a hundred bucks.

    The increase in computational power per dollar in the last fifteen years has been absolutely staggering, and I'm sure that is a big part of the reason why modellers are starting to sound better. I suspect there would have been even more progress if the economic collapse hadn't wiped out manufacturers R&D budgets for several years.

    -Gnobuddy
     

  4. Gnobuddy

    Gnobuddy Friend of Leo's

    Sep 15, 2010
    British Columbia
    Interesting. I haven't tried any of Yamaha's guitar-related electronic toys. But the company has a long and impressive history with DSP - they do a lot of fancy tricks to get their digital keyboards to sound as good as they do.

    For instance I recently bought a budget-priced $250 PSR-E433 - and found out from the manual that Yamaha is actually using DSP to compensate for the poor frequency response of the tiny built-in speakers and make them sound much better than they actually are. Basically Yamaha actually *is* getting a quart's worth of bass into a pint-pot, if I may mix my metaphors a little.

    I wasn't expecting that level of technical sophistication in an entry-level all-plastic keyboard!

    Yamaha doesn't advertise this, by the way, I only found out because the manual tells you to switch to a different internal EQ setting if you intend to use the keyboards line-out signal rather than listen to the built-in speakers. From that I inferred that they were in fact using some sort of EQ compensation curve to make the little speakers sound better the rest of the time, when you weren't feeding the line-out signal to an external amp or mixer.
    Perhaps Line 6's awful-sounding early offerings have attached a permanent stigma to the term "modelling" as far as guitar amps go?

    -Gnobuddy
     

  5. TieDyedDevil

    TieDyedDevil Tele-Holic

    720
    Sep 8, 2006
    Portland, OR
    Later in the paper there are sketches of more sophisticated approaches, such as "dynamic convolution" and "circuit simulation". Something akin to dynamic convolution is probably used in the Kemper Profiling Amp. Circuit simulation is probably used in the Axe-FX and Eleven Rack.

    Some of the other approaches *are* pretty simplistic.

    I suspect that the technique for piecewise-linear approximation originated with analog computing, which probably dates back to the `50s. I remember having been exposed to the technique in the early `70s.

    I know that the Fender Cyber-Deluxe used this technique. I think it may be the basis of Peavey's Transtube.
     

  6. TieDyedDevil

    TieDyedDevil Tele-Holic

    720
    Sep 8, 2006
    Portland, OR
    I'm sure there are still some players who still conflate "modeling" with "POD".
     

  7. Donelson

    Donelson Tele-Afflicted

    May 31, 2011
    Nashville TN
    Nit-picky dept.:

    GarageBand came out in 2004. The modelling stuff predates this by several years. I used Amp Farm in a couple of NJ studios in the late 90's, then tried a Pod. It was so convenient for recording I bought the whole rig myself, with the footswitch unit, in 2000. I even used it on live gigs a few times back then. Too much fiddling around involved so I retired it, then sold it all off. All that was several years before GarageBand debuted.

    Now many years later I have a Digitech RP255, and a Cube 40xl; both have the "modelling" feature.

    To me, there is a phoniness to this technology that reveals itself slowly, after many hours or even months of using it. Very subtle; not perceivable by a listener, only the player. Not unlike a digipiano, such as Ivory et. al. Sounds good, but "feels" weird sometimes.

    Certainly useful, and "better than nothing", but also a compromise that "some" will not like, ever. I'm not that hardcore; sort of in the middle!
     

  8. Gnobuddy

    Gnobuddy Friend of Leo's

    Sep 15, 2010
    British Columbia
    That would be the kindest way to explain to myself the fact that I paid good money for *three* separate Line 6 products over the space of several years. :oops:

    The harshest way, of course, would be to call myself a total idiot who can't even recognize the obvious even when hit over the head repeatedly with it. :eek:

    Each time I convinced myself that if I tinkered with the bloody thing long enough I would get a good sound out of it. Each time I proved myself wrong, but it took a while for the self-denial to dissipate and for acceptance to set in.

    And each time I figured the next product would surely do the trick. :rolleyes:

    If three awful-sounding Line 6 products wasn't bad enough, somewhere along the way I also went out and bought a Digitech RP150. Yay, another hundred different kazoo sounds in one box! :rolleyes:

    So I'm one of the thousands of musicians who've been burned repeatedly by awful-sounding modelling products that didn't come anywhere close to what they promised to be.

    I still have hope for the future, though, as the quality of the models continues to get better, and the power of affordable computer hardware to run the models on continues to increase.

    My most recent venture into digitally modelled guitar pedals have been a Zoom B3 (for bass guitar) and a Zoom G3 (for regular six-string guitar). I have basically no complaints about the B3's sound quality. I haven't used it extensively, but to my ears, it sounds pretty much acceptable.

    I think it helps that our ears have less sensitivity to the lower frequencies, like those in the bass guitars range. Certainly my ears are a lot less tuned-in to the subtleties of bass guitar tone compared to regular guitar tone.

    The G3 is a different story, as all the amp models still sound harsh to me. To be fair, the G3 also contains plenty of digital models that sound good to me - things like reverbs and delays and choruses and so on. It's only the tube amp models that fall on their faces, just like every other tube amp model I've ever heard.

    There is an upcoming Zoom A3 (for acoustic guitar). I found one long demo of the A3 on YouTube, and it seems to sound pretty good. Of course it doesn't attempt to mimic a tube amp, either - all the internal models are either EQ (mimicking the sounds of various acoustic guitars by emulating their body resonances), or time-based effects like reverb and delay and chorus.

    The conclusion to me is that todays affordable DSP technology is very good at doing linear things (reverb, delay, chorus, flanging, tremolo, etc). Unfortunately it still has a long way to go in accurately replicating nonlinear things (like tube amp distortion).

    -Gnobuddy
     

  9. Gnobuddy

    Gnobuddy Friend of Leo's

    Sep 15, 2010
    British Columbia
    Don't forget the Fender Cyber-Twin. Wikipedia says it was launched in January 2001. It combined a digital model running on computer (i.e. DSP) hardware with old-fashioned tubes.

    And how long has Pro Tools been doing its thing? Wikipedia says it started out in 1984 as Sound Designer, which was originally designed to edit sounds for the E-MU Emulator sampling keyboard. In other words it was part of a computer model of an audio device - a sampling keyboard, not a guitar, but that is only a detail.

    Apple may have been one of the earlier computer manufacturers to support multimedia, but their entry into digital audio showed up quite late to the playing field.

    -Gnobuddy
     

  10. 1969spacecowboy

    1969spacecowboy TDPRI Member

    5
    May 4, 2013
    north west
    ignore me!
     
    Last edited: May 5, 2013

  11. dunedindragon

    dunedindragon NEW MEMBER!

    1
    May 30, 2013
    dunedin, fl
    I personally still use the term modeling and have no aversion to it. After all, it is descriptive of what the technology does. I think the aversion to it by some people has more to do with them not understanding the technology and the comfort level they have with older technologies that require less of them. Modeling, whether it's applied to guitar amps, pianos, or drums is nothing more than identifying the physics of sound and replicating those behaviors using digital logic. Heck, we can model an entire universe with a high degree of accuracy, so the behaviors involved in a given amplifier's sound is relatively simple with enough computing horsepower, which is easy to come by these days.

    However, modeling is a more complex process and requires more of both the end user and the manufacturer to feed in the appropriate parameters, or presets. With that complexity you get flexibility you wouldn't otherwise have, and that's where the misunderstanding comes in. I liken it to the days in the late 80's and early 90's when mainframe computer programmers regarded PC's, LAN's, and the Internet as great little toys, but would never be able to do the serious jobs mainframes do. Of course, all of those folks are no longer working in the industry.

    The music industry has been invaded by the same type of advanced technology which (on the surface) appears to be more complex than necessary to those that are used to plugging into an amp, or setting up a drum kit, and playing with the same sound all night long at a gig. But modeling changes that paradigm. If you spend the time to set up things correctly and use the technology the way it should be used, you can accomplish things that would be impractical in the analog world. The simple example is a set of 10 songs, a couple of metal numbers, some country, some blues, some jazzy sounds, and maybe acoustic numbers. To really achieve the "correct" sound you would need to change the amps and the drum kits in an analog world, but that's not the case in the digital world with a modeling amp and an electronic drum kit.

    What amuses me most in these types of discussions is how people talk about achieving the tone they want, without giving any thought to their audience...who can't tell a fender from a marshall, but they know when a song sounds authentic and when it doesn't. The price of that versatility is in the manhours spent planning and preparing it..in exactly the same way as the mainframe versus distributed architecture of computers. If you just want to show up at a gig, plug in, set the amp and play all night, the value of modeling is lost. However, if you want to have each song in your set to have the sound unique to how it should sound, then you'll have to invest the time or money getting the right presets so you can do something that would be virtually impossible in an analog setup.
     

  12. tukk04

    tukk04 Tele-Meister

    150
    Jun 11, 2012
    Stroudsburg, Pa
    It's not a definite divide. There are analog modeling amps that actually switch between circuits (tone stacks, preamp styles, gain structuring, power amp structure) using physical relays and use DSP only for effects. The Fender cyber amps (I regularly use a cyber twin SE (the later one with the tube preamp) for band practice and shows and get compliments on my tone often) use this topology and I believe Marshall now has a mid-level modeling head using the same concept.

    As I've said on other threads, who cares if the modeling is a perfect copy of an amp if it sounds good.
     

  13. Jefe

    Jefe Tele-Holic

    999
    Jun 17, 2004
    Wallingford, CT
    Interesting. I guess if it's using relays and keeping everything alalog, then well, that's solid state. But DSP for effects makes it a digital amp in my book. Or maybe a hybrid.
     

  14. tukk04

    tukk04 Tele-Meister

    150
    Jun 11, 2012
    Stroudsburg, Pa
    The Cyber-Twin SE is basically the best mix of tube, solid state, and DSP in one.
    Tube preamp, analog circuitry, solid state power amp for reliable volume, and at the time state of the art DSP effects with analog overdrive and fuzz. They were ahead of their time when they came out, but their downfall was how much they had to cost to make up for all the work that went into fitting all that goodness into a single amp and making it blend well.

    You can find them for $6-800 now, and they are well worth a try if you come across one, the presets suck but its strength lies in its ability to be customized to your liking. Being a fender, it especially excels at blackface and tweed sounds, I've found it does a mean hiwatt too.
     

  15. Mike Simpson

    Mike Simpson Doctor of Teleocity

    Age:
    61
    Mar 19, 2006
    Gilbert, AZ (PHX)
    No solid state amp with a 12" speaker will ever sound like a Tweed Pro with a 15" speaker or a Super Reverb with 4x10's... at least not to someone that has owned and played a Tweed Pro with a 15" speaker or a Super Reverb with 4x10's...
     

  16. Jefe

    Jefe Tele-Holic

    999
    Jun 17, 2004
    Wallingford, CT
    It's probably better to compare 12" to 12", no? I mean, when making comparisons between types, it's better to keep all other things equal.
     

  17. Frodebro

    Frodebro Doctor of Teleocity Ad Free Member

    Age:
    47
    Aug 17, 2012
    Seattle

    If all you need is a Tweed Pro with a 15, then get a Tweed Pro with a 15. But if you are in a situation that requires a bunch of different amp types, modelers sure are hard to beat.
     

  18. tukk04

    tukk04 Tele-Meister

    150
    Jun 11, 2012
    Stroudsburg, Pa
    If that was directed at my comment, I said "excels" not "exactly replicates" It's not perfect, but the tweed deluxe model has a great crunch and is my main rhythm dirt.
     

  19. Mike Simpson

    Mike Simpson Doctor of Teleocity

    Age:
    61
    Mar 19, 2006
    Gilbert, AZ (PHX)
    I have 2 Tweed Pro Amps.

    You miss the point that modeling does not REALLY sound like other amps.
     

  20. Revv23

    Revv23 Friend of Leo's

    Oct 9, 2009
    Michigan
    Which is why you shouldnt expect them too. They can however, sound quite good. And im a tube slappy

    Sent from my iPhone using TDPRI
     

IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.