Profiling method question​

  • Hi everyone,


    Before I begin, I just want to point out that I have no idea how anything works, have zero audio engineering or programming experience, and am 99.99% ignorant of most things in the universe.


    So from my understanding, the Kemper creates profiles by sending a series of test signals to the amp for it to interpret, then science happens, and then out pops a profile. As we all know, one of the limits to this process is that the Kemper doesn't learn how the reference amp's EQ controls work or interact other than how they're set while the amp is profiled. Same goes for gain.


    But what if the Kemper could learn those things.


    The idea I have in mind, is that the Kemper would send out a test signal targeting a specific frequency range, and then you would sweep the EQ knob for that frequency on the reference amp from it's minimum to maximum for the Kemper to interpret. For example, let's say you have an amp with your standard bass-mid-treb controls. You would tell the Kemper to target those frequencies individually, at which point it would send out the appropriate signal and prompt you to start sweeping the range of the EQ knob. This process would be repeated for each EQ control. Gain could be profiled in a similar fashion.


    So basically, it would be amp modeling by way of profiling, if that makes any sense. Of course, if this doesn't make sense, remember that I know nothing!

  • I outlined an almost-exact copy of what you've proposed a couple o' years ago, and I'm sure others have thought of it too, mate.


    Only difference was I suggested the Kemper take snapshots and extrapolate between them. One would choose the resolution at the start of the Profiling process. IOW, do you capture a knob at lowest and highest setting only, or with more (selectable number) steps?


    Just as you suggested, the Kemper would prompt you to make adjustments throughout the process.


    Same goes for gain.

    Not quite. The gain knob is captured reasonably-well at all settings up to and including the current position on the amp. This is, I've always guessed, because of the varying level of the Profiling signal. So, reducing gain on the KPA provides pretty-good results, and increasing it, especially a lot, takes you into the land of extrapolation without an end-game reference, so results, whilst pleasing, could be way off "reality".


    So, to answer your question, yes, this does make sense, at least to me.

  • Intriguing thought! :-)


    However, even if the Kemper as able to do this, I don't think it would work as described - consider, for example the bass and treble knobs on something like an old ac30. Those knobs are highly interactive, and without sweeping both knobs at the same time, the Kemper would probably learn something "wrong" and not give an accurate result.

  • That sounds more like a modeller really. I look at the Kemper as saving a sound I really like from a setup and then it can be tweaked by the controls I have come to understand. It never occurred to me to try and replicate every feature in the amp profiled.

  • If it could work I suppose it would eliminate the need to profile loads of settings of a given reference amp? Basically you would only have one profile per amp and you could tweak to however you want?

  • That sounds more like a modeller really. I look at the Kemper as saving a sound I really like from a setup and then it can be tweaked by the controls I have come to understand. It never occurred to me to try and replicate every feature in the amp profiled.

    No mate, as @Deej09 said, it'd simply be a way of combining multiple snapshots instead of just employing a single one in a Rig. Modelling is a different kettle of fish altogether.


    Capturing the interactivity of EQ bands is IMHO something that'd have to wait - it'd logically be further down the development line than the initial move to the combining of multiple Profiles.

  • Without capturing the interaction between the tone controls, I don't think it would be any better tan the current Kemper EQ block.


    Not only are the EQ controls on valve amps interactive but they also affect gain in many cases and this varies depending on whether they are pre or post gain stage (Meas Boogie Mark series etc) so capturing them properly in a profile would require some serious voodoo. Mind you people would probably have said the same about profiling snapshots of real amps before CK did it ;-)

  • No mate, as @Deej09 said, it'd simply be a way of combining multiple snapshots instead of just employing a single one in a Rig. Modelling is a different kettle of fish altogether.
    Capturing the interactivity of EQ bands is IMHO something that'd have to wait - it'd logically be further down the development line than the initial move to the combining of multiple Profiles.

    I think we all know that the profiler is probably about 85-95% accurate. If you rely on profiling many elements, that would simply multiply the inaccuracies, should you want to be true to the original amplifier. Creating interaction between the tone controls would be difficult at best, because some have an impact on gain. My preference would be to improve the basic profile before any other bells and whistles, if that is possible.


    When I bought the Kemper, it was all about being able to recreate a multitude of different products, but soon I started to think of the Kemper as a unique amp and only use profiles as a starting point. Using the EQ, it I don't think was this the way a Mesa treble control would work; I simply want more or less top end and then start playing. In any case, many amplifiers sound very similar at high gain and tone is more to do with the cabinet from what I hear.


    The main problem moving forward for Kemper is that the users all seem to have different needs and wishes. Maybe the best thing is for them to forge ahead with their own agenda, as that was successful with the original product.