About the dynamic compression/expansion
That's been a long time I wanted to check what happens with dynamic compression or expansion
How would that measure ?
So I've build a digital loopback with Cockos Reaper and Virtins Multi-Instrument 3.9 (the software I use for audio measurements)
Then I added some compressor and gate in the loop
Here are the settings
The compressor
The gate
Then I compensated the level in Reaper's mixer by +11.3dB
And I measured THD vs level, IMD vs level and, of course, the output level
I've combined all those values in the same plot, to explain
What we can see
1. With Reaper's input level (dashed green), we see that Reaper's output level (dashed black) is influenced by the gate (below -30dB) or by the compressor (above -12dB)
Between those threshold, influence is null (you just see my final level compensation)
2. IMD (blue) and THD (red) have a normal decreasing plot.
This is similar to what you see in Amir's normal review
(If you except that, here, we have everything in digital 24 bits domain, so we have a VERY low THD and IMD)
But appart a few incidents in the gate domain (probably due to window averaging), we see little clue of any effect.
(NB: It's normal that the noise floor appears higher for IMD, since that's relative to the 7kHz signal, which is 12dB lower than the 1kHz signal used for THD)
3. Noise floor (which I represented in orange as dBFS, so not related to signal, like in SNR), varies with compression.
That's to be expected.
For completeness, I added THD+N (-SINAD) (yellow)
The gate is visible here.
It is in fact a brutal level change, not a smooth expander.
So, if I'm not mistaken, for a normal, physical, device,
you would only see that there is some dynamic compression in the level linarity measurement
(Only if the compression is active at the frequency used for that test. But if it is frequency dependent, you'll probably see it in frequency response)
But I can assure you that this kind of treatment is VERY audible.
@amirm
Is that correct ?
Or is there another test that will show this effect is at work ?
(I originally posted this in the Asus noise reduction dongle review.
I think it makes more sense here)
That's been a long time I wanted to check what happens with dynamic compression or expansion
How would that measure ?
So I've build a digital loopback with Cockos Reaper and Virtins Multi-Instrument 3.9 (the software I use for audio measurements)
Then I added some compressor and gate in the loop
Here are the settings
The compressor
The gate
Then I compensated the level in Reaper's mixer by +11.3dB
And I measured THD vs level, IMD vs level and, of course, the output level
I've combined all those values in the same plot, to explain
What we can see
1. With Reaper's input level (dashed green), we see that Reaper's output level (dashed black) is influenced by the gate (below -30dB) or by the compressor (above -12dB)
Between those threshold, influence is null (you just see my final level compensation)
2. IMD (blue) and THD (red) have a normal decreasing plot.
This is similar to what you see in Amir's normal review
(If you except that, here, we have everything in digital 24 bits domain, so we have a VERY low THD and IMD)
But appart a few incidents in the gate domain (probably due to window averaging), we see little clue of any effect.
(NB: It's normal that the noise floor appears higher for IMD, since that's relative to the 7kHz signal, which is 12dB lower than the 1kHz signal used for THD)
3. Noise floor (which I represented in orange as dBFS, so not related to signal, like in SNR), varies with compression.
That's to be expected.
For completeness, I added THD+N (-SINAD) (yellow)
The gate is visible here.
It is in fact a brutal level change, not a smooth expander.
So, if I'm not mistaken, for a normal, physical, device,
you would only see that there is some dynamic compression in the level linarity measurement
(Only if the compression is active at the frequency used for that test. But if it is frequency dependent, you'll probably see it in frequency response)
But I can assure you that this kind of treatment is VERY audible.
@amirm
Is that correct ?
Or is there another test that will show this effect is at work ?
(I originally posted this in the Asus noise reduction dongle review.
I think it makes more sense here)
Last edited:
