An MIR analysis of real amp versus captures and profiles

AmpliTube TONEX, AmpliTube 5, AmpliTube Custom Shop, AmpliTube Joe Satriani, AmpliTube Brian May, MESA/Boogie®,Orange™, Fender™, Hendrix™, Metal, AmpliTube SVX, and more for Mac/PC

An MIR analysis of real amp versus captures and profiles

Postby avernon1 » Sun Oct 30, 2022 4:13 pm

I did some messing around with Python and made some graphs. Check it out here:
https://medium.com/@acolmiztli/an-mir-a ... 5d4d415535
avernon1
 
Posts: 11
Joined: Sun Mar 13, 2011 1:05 am

Re: An MIR analysis of real amp versus captures and profiles

Postby dean.winter » Sun Oct 30, 2022 5:57 pm

Wow, that is super cool. I really liked that analysis, particularly looking at the mel stuff and comparing the 'fast' vs 'advanced' capture modes. Out of curiosity, did the mel scale differences shown in the graph line up with what your ears were telling you for the Kemper/QC stuff? I always think it's cool to see something graphed that makes me say "oh yeah, that's what my ears were hearing"

Dean
dean.winter
 
Posts: 44
Joined: Mon Apr 11, 2011 4:20 pm
Location: SLC, UT

Re: An MIR analysis of real amp versus captures and profiles

Postby avernon1 » Sun Oct 30, 2022 10:48 pm

Before the last update, I could detect some high frequency loss. But with this most recent update, I'm having a lot of trouble picking out the differences - aside from noise floor.

But yes, the findings did clarify what I hear. With the Kemper I hear a fuzzy low-end, with the emphasis in the wrong place. That is what I see in Fabfilter Q3 when I compare against the real amp, and you can see all of the low-end activity in the Kemper tracks that is extra data that wasn't there in the original amp.
avernon1
 
Posts: 11
Joined: Sun Mar 13, 2011 1:05 am

Re: An MIR analysis of real amp versus captures and profiles

Postby RiF_2205 » Mon Oct 31, 2022 7:44 am

Extremely interesting. this could help settle down those endless 'but I am hearing / feeling this and that' discussions. And as dean.winter said, if you hear / feel something and then can scientifically prove it, that's fine. If you hear something but there's no proof, it's probably purely psychological (like sometimes my amps just don't sound right, but on the next day they do).
And I learned something about the Mel scale, Fourier Transform and how to read spectograms on the way.
User avatar
RiF_2205
 
Posts: 57
Joined: Fri Dec 28, 2012 7:19 am

Re: An MIR analysis of real amp versus captures and profiles

Postby thomas.brunkard » Tue Nov 01, 2022 1:30 pm

avernon1 wrote:I did some messing around with Python and made some graphs. Check it out here:
https://medium.com/@acolmiztli/an-mir-a ... 5d4d415535



Great work and a fantastic post.
Check out my adventures with AmpliTube Brian May and Axe I/O here: https://www.youtube.com/playlist?list=PLzJ0af-klLVmUizM3dqL8g-oFwEYBTxAu
thomas.brunkard
 
Posts: 49
Joined: Mon Feb 03, 2020 3:05 pm
Location: Dublin, Ireland

Re: An MIR analysis of real amp versus captures and profiles

Postby blackemm » Tue Nov 01, 2022 6:18 pm

Fantastic stuff!
blackemm
 
Posts: 15
Joined: Fri Jun 05, 2020 1:06 am

Re: An MIR analysis of real amp versus captures and profiles

Postby avernon1 » Tue Nov 01, 2022 7:48 pm

Cheers. I thought I might do some other experiments if I get some time. I don't have a Quad Cortex or Kemper anymore. But I thought I might try doing something with my Axe FX 3 too. Just to see what the differences are between the amps in there that match up with my real-world amps, and ToneX.
avernon1
 
Posts: 11
Joined: Sun Mar 13, 2011 1:05 am

Re: An MIR analysis of real amp versus captures and profiles

Postby blackemm » Wed Nov 02, 2022 1:42 am

Interesting to quantify the differences, but I'm not sure it would be a fair comparison. (Control)

Were your graphs based on direct captures? (Eliminate Mic, IR, etc.) If so? VERY interesting. I know it "feels" like an amp, to me, but I DO like any evidence that may/may not help quantify what I'm hearing.

Well done.
blackemm
 
Posts: 15
Joined: Fri Jun 05, 2020 1:06 am

Re: An MIR analysis of real amp versus captures and profiles

Postby avernon1 » Wed Nov 02, 2022 9:15 am

blackemm wrote:Interesting to quantify the differences, but I'm not sure it would be a fair comparison. (Control)

Were your graphs based on direct captures? (Eliminate Mic, IR, etc.) If so? VERY interesting. I know it "feels" like an amp, to me, but I DO like any evidence that may/may not help quantify what I'm hearing.

Well done.


Direct captures with IR's applied.

I think feel is downstream from frequency response personally. Different "feel", then you'll have a different frequency response, and vice versa.
avernon1
 
Posts: 11
Joined: Sun Mar 13, 2011 1:05 am

Re: An MIR analysis of real amp versus captures and profiles

Postby blackemm » Wed Nov 02, 2022 2:26 pm

Direct capture, same IR makes for a good comparison! . Wish you still had the other's. It'd be interesting to compare a handful of different amp profiles.

EQ is a good starting point. I really don't know how one would go about "scientifically" testing "feel." I wonder if tonex is feeling good because its closer to modeling the non linearity of the amp? Is it that "chaos" its trying to learn in AI? (After EQ) And how would one graph that?

I like the reasons why. Doesn't mean squat to most. They'll like what they want(or what they are familiar with). I know my captures sound like my amp. Your graphs may have quantified why. But it also feels like my amp. THAT has been my beef with the digital stuff, all along. Is it just that my ears WANT my amp, and that familiarity? Or is there something else that can bw quantified in the response?
blackemm
 
Posts: 15
Joined: Fri Jun 05, 2020 1:06 am

Re: An MIR analysis of real amp versus captures and profiles

Postby blackemm » Wed Nov 02, 2022 2:27 pm

*duplicate
blackemm
 
Posts: 15
Joined: Fri Jun 05, 2020 1:06 am


Return to AmpliTube Guitar Amp & FX modeling