Speaker Simulation Test: Q-Clone Vs. Convolution Reverb

Discussion in 'Recording/Live Sound' started by Orren, Apr 28, 2005.


  1. Orren

    Orren Member

    Messages:
    1,143
    Joined:
    May 17, 2004
    Location:
    Orange County, CA
    Ed DeGenaro inspired me! :)

    I had used convolution reverbs to emulate speakers in the past, but Ed's Q-Clone files got me wondering how the Q-Clone algorithms compare to more traditional impulse response algorithms used in convolution reverbs.

    So I did an A/B/C test, trying to be as fair about it as I could. The process was:

    1) I played a short (24 second) guitar riff through the Flexi-50 amp. The full signal chain was Guitar-->Flexi-->Hot Plate-->2x12 Speaker Cabinet. The riff was simultaneously recorded into Logic onto two seperate mono tracks. The first track was a Sennheiser e609 close on a V30. The second track was a direct signal from the line out of the Flexi into the Line In of the Fireface.

    --> The mp3 of this track is located at: http://www.mertonfolio.com/Flexi+V30 Speaker Cabinet.mp3

    2) I set up the "Q-Capture" Plug-in on a track. That track was set to output directly to one of the line outs, and straight into the Flexi power amp section (set clean, with full treble). I used the same mic, and recorded the same speaker, as above. As soon as I hit record, the "Q-Clone" plug-in then captured the EQ curve of the Speaker. I then applied the newly made QClone preset to the Flexi direct track.

    --> The mp3 of this track is located at: http://www.mertonfolio.com/Flexi+QClone.mp3

    3) I grabbed the Sine Sweep IR I use to make simulations for Space Designer the convolution reverb included with Logic Pro, and recorded the speaker sine sweep. I used the same mic, and recorded the same speaker, as above. The I used the Deconvolution process to make an IR from that. Finally, I trimmed it in the Sample Editor, so it was usable. I then applied the newly made SDIR to the Flexi direct track.

    --> The mp3 of this track is http://www.mertonfolio.com/Flexi+SD.mp3"

    Here are the Match EQ curves of the 1) mic'd speaker:
    [​IMG]

    2) Direct+Qclone
    [​IMG]

    and 3) Direct+SD
    [​IMG]

    Clearly, they all look pretty close. I also used the "match" button" to see how the "learned' simulation signal would need to be adjusted to match the "template" original signal. 1) is the QClone Match
    [​IMG]

    2) is the SD Match
    [​IMG]

    Its interesting that the guitar signal itself contained almost no information below around 80k or above 5k,and both simulations seemed to "do their own thing" beyond those extremes. It looks like SD was more accurate by 1 or 2 dB until about 2kHz, at which point the SDIR seems more accurate in general.

    I think the bottom line is how both of them, while not exactly like the speaker, sound VERY close, and using software it is possible to mic a speaker one time, and then re-apply that speaker's response curve to other, direct mic'd guitar tracks. This of course means that you can capture those raging guitar tones without being super loud!

    Hope this was useful to someone, at least! ;)

    Orren
     
  2. TAVD

    TAVD Guitar Player Gold Supporting Member

    Messages:
    3,429
    Joined:
    Jan 20, 2004
    Location:
    MD
    The first mp3 link is broken.
     
  3. tms13pin

    tms13pin Supporting Member

    Messages:
    1,592
    Joined:
    Feb 26, 2005
    Location:
    Pittsburgh, PA
    > Its interesting that the guitar signal itself contained almost no > information below around 80k or above 5k,and both
    > simulations seemed to "do their own thing" beyond those
    > extremes. It looks like SD was more accurate by 1 or 2 dB until
    > about 2kHz, at which point the SDIR seems more accurate in
    > general.

    Don't have speakers or headphones on my PC here at work, so
    I haven't listened to the files yet. But, the low E on a guitar is
    at about 83 Hz, and guitar cabs roll off at fairly low frequencies,
    so its not surprising at all that your BW here is between 80 Hz
    and 5 kHz!

    --Tom
     
  4. Orren

    Orren Member

    Messages:
    1,143
    Joined:
    May 17, 2004
    Location:
    Orange County, CA
    Actually, it's not so much "broken" as I named the file badly. :) If you copy the link--including the spaces, since the words "speaker cabinet" don't connect--it will work just fine. I tried it right now, in fact, and it works.

    Sorry about that (I don't want to rename the file, since I liked to it elsewhere), but copy + paste isn't that bad.

    Orren
     
  5. renico00

    renico00 Member

    Messages:
    238
    Joined:
    Jul 24, 2002
  6. wilder

    wilder Member

    Messages:
    769
    Joined:
    Feb 20, 2004
    Location:
    Denton, TX
    I'm going to have to try this at home.

    Thanks!

    Chris
     
  7. MichaelK

    MichaelK Member

    Messages:
    6,477
    Joined:
    Jan 7, 2002
    Location:
    Fort Mudge
    Still, we must have a monstrously expensive preamp, musn't we? :D

    Just kidding... didn't mean to hijack. But that lack of bandwidth raises an interesting point. With so little to work with the "match" graphs could be misleading.

    Say the signal strength at 18k was -140 dB. A +/-10 dB difference in response is the difference between -130 and -150! In other words, The signal level at those frequencies that showed a significant difference was so microscopic that a comparison might not be a valid.

    I would be interested to see the differences on a wider bandwidth sample, like a piano or acoustic guitar, or a few different instruments.

    Interesting stuff, Orren. Thanks.
     
  8. MichaelK

    MichaelK Member

    Messages:
    6,477
    Joined:
    Jan 7, 2002
    Location:
    Fort Mudge

Share This Page