When the Noise Floor on Compressor is set to max (-20db), all audio experiences negative gain. Why?

The question hasn’t changed. I’m not asking how to fix anything. This was an experiment to begin with.

  • So, should sounds below the Noise Floor be reduced or unaffected? (It looks like you’ve said both) I thought any sound below the Noise Floor should be ignored and unaffected. Compression “based on peaks” always uses upward compression in Audacity regardless of whether or not you select “make-up gain,” so I could see that causing the compressor to be overly engaged, but in my case, I’m using RMS, not peaks.

  • How would the attack/release times create this issue? What’s so special about them being short? The ambient noise lasts for 7s, giving the compressor plenty of time to disengage, no?

  • You said the algorithm could cause compression to overlap or applied more broadly if it’s near the Noise Floor or Threshold. How so? Are they not strict limits? This must be what’s happening, but again, 7s of audio beneath the Noise Floor should give the compressor plenty of time to disengage.

Thanks, DigiGod!

I don’t normally use compression and I don’t know anything about the algorithm so feel free to ignore what I say. :stuck_out_tongue:

That sounds like a general definition of dynamic compression. Most compressors “push down” the loud parts and sounds below the threshold are not changed (ignoring the knee around the threshold) unless/until you use make-up gain, which brings everything up linearly after compression.

With make-up gain, the compressor is still working on the loud parts, but the end result is the opposite - bringing-up everything below the threshold, including the background noise.

Most often, compression is used to make “everything louder”, so make-up gain is usually part of the process, even if it’s done separately after compression.

Since compression reduces the dynamic range it reduces the signal-to-noise ratio and the noise becomes more noticeable, especially with make-up gain which turns it up along with everything else.

So my assumption is that the noise floor setting applies downward dynamic expansion, lowering the volume at low levels to offset (or reverse) the increase in background noise.

…GoldWave has in interesting Compressor/Expander. There are compression presets to “reduce loud parts” above the threshold (downward compression like a normal compressor) or to "boost quiet parts below the threshold (upward compression).

The expander has settings to “boost loud parts”(upward expansion) or to "reduce quiet parts (downward expansion). (GoldWave actually fouls-up the definitions of compression & expansion but if you start with the presets it does what it says.)

Using audacity’s compressor (which is the only one I’m asking about), there is a difference between using RMS or “based on peaks.” According to Audacity’s wiki:
Noise Floor: The compressor adjusts the gain on audio below this background level so as to prevent it being unduly amplified in processing.

In other words, things below the Noise Floor should not be amplified by the compressor. I’m using RMS (not “based on peaks”), and I have not selected to “make-up gain,” so nothing below my chosen Threshold should be amplified. In my case, it’s actually experience negative amplification. Still trying to figure out why.

So my assumption is that the noise floor setting applies downward dynamic expansion, lowering the volume at low levels to offset (or reverse) the increase in background noise.

You might be on to something here. Perhaps all of my audio is being compressed and then being corrected afterward for audio below the Noise Floor, but in my case, somehow, it’s being overcorrected, causing audio below my Noise Floor to be downlardly compressed rather than unchanged. The fact that Audacity’s definition stated “compressor adjusts the gain… to prevent it being unduly amplified in processing,” to me, implies that it must all be compressed first, otherwise it wouldn’t need to be adjusted.

Thanks, DVDdoug.