Interesting article on testing Compressor response characteristics

Author
Post
Essentials Only Full Version
The Maillard Reaction
Max Output Level: 0 dBFS
  • Total Posts : 31918
  • Joined: 2004/07/09 20:02:20
  • Status: offline
2009/10/30 22:43:51 (permalink)

Interesting article on testing Compressor response characteristics

Interesting article on testing Compressor response characteristics:

http://www.garage27.com/f...fler_masterProject.pdf

here's an excerpt I found interesting:

"Test tones
As already explained, a critical part of the comparison is, to keep the settings
of all test-processors as equal as possible, a method of actually measuring
important parameters has to be developed.
The initial idea was to use white-noise test signals, to measure attack,
release and gain reduction in ProTools or Logic’s sample editor. But after
several attempts, the conclusion was drawn that this method is unusable
because of the more or less random waveform of a white-noise signal, as
seen in picture 3.1
Especially when it comes down to really fast time constants, this method
is inappropriate, which led to the following, quite obvious improvement:
Regular sine-waves were used instead, as they deliver a much more usable
result. After experimenting with different frequencies, a 3kHz tone has been
chosen, since it offers a short enough wavelength to be useful within a 5ms
range, whilst still providing a useful signal for the compressor to react to.
Experiments with different frequencies (100, 1k and 10kHz) will show, if
there are differences in the detection circuits.
Gain reduction can be measured with the same test tone principle and
a simple level meter, which is the last important setting that needs to be
measured."

I didn't have the attention span to read it all this evening, I'm curious to learn what other people think of the test procedure etc.


best regards,
mike




#1

3 Replies Related Threads
    drewfx1
    Max Output Level: -9.5 dBFS
    • Total Posts : 6585
    • Joined: 2008/08/04 16:19:11
    • Status: offline
    Re:Interesting article on testing Compressor response characteristics 2009/10/31 10:58:37 (permalink)
    Interesting analysis. I thought it was very well done, though it only went so far (I mean that as an observation, not a criticism).

    Some additional things that could be investigated more thoroughly:

    1. Zero compression performance - how is each coloring the sound when in non-compression mode. Distortion, noise, frequency reponse, phase.

    2. Low frequencies/fastest attack+release - since this is a (very fast) feedback compressor, the compressor tracks/modulates/distorts low frequency waveforms. "Formally" testing this might reveal the relationship between the input level and the preset thresholds of the feedback control section. And since the compresser is going through attack/release so rapidly (with each cycle), differences in attack/release might be magnified to produce very different output waveforms on different implementations. Of course, since you are distorting the waveform here (i.e. adding high frequency harmonics), you should run the SW versions at the highest possible sampling rate and make sure the HW versions are band limited to match before comparing the resulting waveforms. 

    drewfx
    #2
    jimmyman
    Max Output Level: -53.5 dBFS
    • Total Posts : 2193
    • Joined: 2008/12/16 06:57:38
    • Status: offline
    Re:Interesting article on testing Compressor response characteristics 2009/10/31 11:34:54 (permalink)

      What I really like in your quote is they say white noise
    didn't work. I prefer a sine wav for running tests. Even
    for EQing a small venue live sound system. Using a sine
    wav generator with a variable pot and being able to
    "sweep" from 20 hz to 20k hz is extremely revealing.

      Using noise for testing is of course also good too but
    it's a different type of test. A pure sine wav will reveal
    so many things such as rise and fall times that may be
    impossible to accurately test with noise.

         

    My soundclick
    #3
    bitflipper
    01100010 01101001 01110100 01100110 01101100 01101
    • Total Posts : 26036
    • Joined: 2006/09/17 11:23:23
    • Location: Everett, WA USA
    • Status: offline
    Re:Interesting article on testing Compressor response characteristics 2009/10/31 13:37:27 (permalink)
    "We are clearly in an age now where high-quality mixes do not necessarily need costly analogue systems..."

    Blasphemer! And not a single mention of magic fairy dust.


    All else is in doubt, so this is the truth I cling to. 

    My Stuff
    #4
    Jump to:
    © 2025 APG vNext Commercial Version 5.1