Kemper Profiler MK 2

  • Thread starter Thread starter Deleted member 490
  • Start date Start date
Anyone can add these post processing controls if they wanted, nothing extremely novel really. NAM universal plugin has some extra stuff right there for cleaning up / tweaking
 
Whether or not they need them, they are not even an option with the Black Box modelling approach. Machine learning makes a numerical model that is opaque even to creators and vendors of that model.
I think my experience is more that Kemper needs them - they’re not so much an optional tonal adjustment so much as a crutch for when things don’t turn out as expected. I would rather Kemper focussed more on nailing the accuracy - the advance parameters were just a highly efficient way to make your tone go weird.

Any black box solution could tack on other features that achieve a similar result. I don’t really like going that direction, but I believe NAM Universal and Genome have some tweaks like that?
 
LOL that is perfect timing of course, a week after I sell the Kemper. Oh well, all the other points still stand.

Let's say they do nail the sound and get in the ballpark of the new QC and Tonex captures...they still have a massive uphill battle to get people to buy the next gen hardware.
 
I don't have the exact analogy for this but being that Marshall stuff is already Kemper's bread and butter generally speaking; posting soundclips of one as "proof of improvement" is laughable.
 
I don't have the exact analogy for this but being that Marshall stuff is already Kemper's bread and butter generally speaking; posting soundclips of one as "proof of improvement" is laughable.

Well I don't hear the signature Kemper "TS808 always on" sound in this clip, but yeah the real test will be something high gain.
 
Be sure to let Steve Atkinson know so that he can ring up Cristophe and tell him he is obsolete.
Huh? Kemper’s accuracy, or lack thereof is on Christophe. If he’s reached the limit of what he’s capable of his own then he can hire any number of professionals to help. If he wants to improve by any significant amount, which I’m assuming he’s not that fussed about.
 
stand by me jerry oconnell GIF
 
Be sure to let Steve Atkinson know so that he can ring up Cristophe and tell him he is obsolete.

Chocol8:​

Why would Steve waste his time with Kemper when there are so many more promising companies with real futures that he can team up with?
_______________________________

So am I getting this right (?) -- there's a whole team of you rockstar guys on here with getting on for 20k posts between you who interpreted my "suggestion" above completely literally. You thought that I believed that this "phone call" would and should happen? And you needed to disabuse me of this fanciful notion? And then make with the (n)(y) stuff?
Fascinating.

Maybe I could bring a whole boatload of irony to where you live and remedy the apparent deficiency - unless it is the USA - bloody tariffs y'know.
 

Chocol8:​

Why would Steve waste his time with Kemper when there are so many more promising companies with real futures that he can team up with?
_______________________________

So am I getting this right (?) -- there's a whole team of you rockstar guys on here with getting on for 20k posts between you who interpreted my "suggestion" above completely literally. You thought that I believed that this "phone call" would and should happen? And you needed to disabuse me of this fanciful notion? And then make with the (n)(y) stuff?
Fascinating.

Maybe I could bring a whole boatload of irony to where you live and remedy the apparent deficiency - unless it is the USA - bloody NO POLITICS y'know.
I have failed to ascertain what it is that you are talking about.
 
Funny how none of the more accurate approaches need these controls (nor see the benefit in adding them)

They mostly can't do it. The more accurate approaches can't change the profile (model) unless it's inherent in their underlying "model". Underneath the covers, Kemper seems to capture key parameters for a model using features that their engineers understand and can tweak. NAM and the like only let you change what happens before and after the model.

There's a long standing tradeoff in the ML world: smaller, simpler models that are easy to understand (and change) ... and dramatically larger, complex models which can be more accurate across a larger set of inputs ... but are effectively black boxes. There are models where a human chooses the features (components of the model) and then there are models where the data (capture) drives the features that have no inherent meaning to a mere human.

I suspect some of the other approaches can see the value of tweaking a capture in ways that make sense to users.
 
Back
Top