It sounds great...but how does it work?

  • I never have seen a thread about how the Kemper produces it great sound. Maybe it is not decent to start such a thread because this procedure is the secret behind KPA's success and I guess it is protected via a patent. But I am very curious what theoretically could explain the differences in sound quality between different digital devices. If the moderator decides to close this thread, I can understand, but I hope we could start an interesting discussion on the principles of digital sound modelling.
    I guess in the early devices in some way the clipping (distortion) proces was digitally reproduced and after that the effect of the speaker and cab mimiced by some form of equalisation of the clipped amp sound. By optimising this clipping and equalising the end result will be of course better, and that is what I think to hear in products like axeFX and now the new helix. But the KPA must have found an other way to achieve the complex way a guitar amplifier distorts the guitar signal, the fingerprint sound from the KPA is really different from the other gear. This strucked me again when hearing the new helix, which imo indeed has the same line6 fingerprint in it as it predecessors. I'm really fairly ignorant in how things are technically worked out in amp modelers and I'm wondering if somebody has an idea how this could be achieved in the KPA.
    I think an important part in a good sounding modeler is phase allineation. When you record and you use inferior equalising plugins you get this 'washed out' sound (like old L6 gear has), more exactly: the opposite from an 'in your face' sound. I guess a speaker will always be phase-coherent because there is no other way for it to move and when you place a mic directly in front of it, you keep this coherence in the recorded signal what gives the 'in your face ' sound. Plug ins in DAW's have evolved over the years and the last years lots of EQ's can be bought which don't influence the phases of different frequencies during the eq-process. These eq's almost always sound better, especially when you want to preserve the directness and openess of the sound. I think these quality improvments in DAW-eq-ing is comparable with the improvements in gear from ie L6. But KPA must have done it in another way: it sounds really 'in your face' and still after five (?) years after introduction is ahead of the competition on this special topic of modelling. That makes me think that the solution must be found in the way the clipping sound is produced. If you manage to form the initial clipping as a function of not only amp characteristics, but speaker/cabinet characteristics as well I think you could have much more control over the endresult. This would be very different from first producing an amp-like clipped signal which you adapt with eq-ing.
    I'm very curious if somebody has an idea how it works, but again, maybe it is not appropriate to discuss such matters on this forum.

  • Sinmix started a thread on his theories of how the KPA works. There were a few others that chipped in with their own ideas, too. It's an interesting read. Cliff Chase from Fractal, the makers of Axe FX, reckons that there are 7 different basic amp models. During the profiling process, the KPA selects the closest model to use as a start point, then does some EQ matching before capturing an IR of the cabinet and microphone. Sounds plausible, but could also be way off!


    Me, I'm not bothered how it does its thing. I'm just glad that it does!

  • It's described in general in the manual in one of the chapters - I think it's the quick start one, not the reference manual.


    If you can read and understand the patent, you could probably deduce at least SOME of the internal workings. But maybe at a more conceptual level rather than detailed. And whatever has been patented is by definition not secret :) Some of the finer points may be, of course.


    I think also during the profiling process you're supposed to start from a pre-existing profile that s"sounds a bit like" the amp you're trying to capture, right? That would back up Sambrox's mention of the x number of "base amp models".


    In the end, it's a matter of having identified the parameters that are necessary to measure, as well as their interaction based on input signal. That input signal is, of course, complex, with regards to attack, decay, sutain and release, as well as volume, harmonics, and simultaneous signals (like two strings together) blah blah blah.... :)

  • That is an interesting document, but hard to understand !http://worldwide.espacenet.com…spacenet.com&locale=en_EP This thing from my first post about phase is not what it is about. If I understand correctly the KPA shifts between two kinds of amplification: one for low input signals which is linear and is affected by the output signal and one for high input signals which is non-linear and is affected by both input and output signal. The switch between these two amplications is non-linear and dependent of the input signal if i understand correctly.
    Why kpa is different from the L6 then must be how the way the amplification is stored into the signal processor, because this kind of global describing of the amplification seems to me not to be patented?

  • It should work, but I see an exclamtion mark stands right in front of it, that isn't part of the link. And I saw a mistake I made: the linear part of the amplification (low input signals) it is affected by the input signal, not output.
    I found another interesting thread: http://allsignalprocessing.com…amplification-of-signals/
    In the end it states the equalisation of a non-linear amplitude signal is the hard part to do because of lack of analytical tools (? I don't understand complete :) ) The invention of Kemper is a solution for this problemen I guess.
    I guess then why a KPA sounds different from a axefx or line6 must be in the wat this non-linear amplification is programmed ? Maybe like Sambrox wrote with 7 different Models?

  • After reading the crux of the patent, I'm leaning more towards Sinmix's version of how the KPA works. Traditional modelling for the Stomps, a super flexible and clean modelling engine for the Amp section and a combined IR and frequency oscillator for the Cab.

  • I think also during the profiling process you're supposed to start from a pre-existing profile that s"sounds a bit like" the amp you're trying to capture, right?


    the profile you start with/have active when you switch to PROFILER mode has no influence on the outcome.


    it is only recommended to start from a profile similar to what you're aiming for to give you a chance to A/B your setup with an existing/'good' tone - page 1/3 of the Profiling Assistant: "Kemper amp" "Reference amp" soft knobs.


    this way adjustments in gain, EQ, mic-ing can be made before the profiling happens.
    that is all. :)


    hth


  • Thanks for the info :)

  • Sinmix started a thread on his theories of how the KPA works. There were a few others that chipped in with their own ideas, too. It's an interesting read. Cliff Chase from Fractal, the makers of Axe FX, reckons that there are 7 different basic amp…


    Hi guys, I may be an idiot, but I cannot find the thread referenced that was started by SinMix talking about the Kemper technology. Can anyone please point me to it?


    Thanks!

  • Quote from nathanmachine: “Quote from sambrox: “Sinmix started a thread on his theories of how the KPA works. There were a few others that chipped in with their own ideas, too. It's an interesting read. Cliff Chase from Fractal, the makers of Axe FX,…


    Thank you! I actually saw that one in my search but discounted it thinking it must be an older thread I was looking for......thanks again.


  • Wow. Thank you, Don.


    I'd heard it said so many times that you're best off choosing a similar rig to what you're going to profile that I thought it was necessary.


    That's great news, mate.


  • Wow. Thank you, Don.


    I'd heard it said so many times that you're best off choosing a similar rig to what you're going to profile that I thought it was necessary.


    That's great news, mate.


    from the Profiling Guide:
    "Start by choosing a Rig in Browse Mode. It’s a good idea to choose a Profile that is somewhat related in character to the reference amp. This will give you the chance to compare the existing Rig with that of the setup you are about to profile, thereby ensuring that you get a result that is even better suited to your needs."


    ;)

  • I love Sinmix's work. I think he has his own sound as a producer and i think his business model is absolutely brilliant -
    but owning a (very big) bunch of his profiles, i can attest to most of them being very similar.
    I can't say that about any other profiling studio, though, and i own several thousand profiles outside of the RE.


    My two notions are as follows;
    1. Technically, a sample-long IR is nothing more than a complex EQ. We already know the IRs on the Kemper Profiler are fairly short.
    If the Profiler used the same dynamics engine for all profiles, an EQ-match should bring different profiles pretty close together - but i've spent a long while trying to get Tim Owens's Splawn to react somewhat like TAF's JCM800, and i've got a good, well-equipped studio at my disposal. No luck, they are dynamically very, very different.
    To prove Sinmix's point, proving you can create similar profiles isn't enough. It's not proof of anything.
    You'd have to disprove the ability to create differing profiles.
    Good luck.


    2. If you told me you could make a catapult out of two matches and a rubber band, i'd have to see it to believe it - but if i saw it, it would make it all the more impressive. If the Profiler can do what it does with a wave-shaper, a bunch of presets and an IR engine - that's brilliant.

    The Axe-FX II is by far a more 'powerful' audio processor than the Kemper. Then again, your studio PC is likely much, much stronger than that, but Amplitube 2 doesn't sound as good as either.
    Specs are never as important as clever algorithms and good choices.

    "But dignity is difficult to maintain
    stamina requires constant upkeep
    repetition is boring
    and you pay for grace."

    Edited once, last by Quitty ().

  • While it's clear to us, I believe this still creates some confusion in the newbie or the casual approacher. It might be worth mentioning in the basic manual.


    :)


    It is probably a bit over the top to anticipate what user could read between the lines of our manual and then negate it in the next sentence :)