compression question

  • Thread starter Thread starter verytorpe
  • Start date Start date
V

verytorpe

New member
hello

if:

i'm always reading that a compressor reduces volume some dbs specified by a determined ratio when the loudness of a sound goes over a determined threshold. so, if i set my compressor threshold to -12 db and its ratio to 3:1, every 3db over that threshold will be squashed to 1 db --> 9 dbs drops to 11dbs (maybe 12?);the volume is reduced

then:

why, when i set a very low threshold, let's say, -40db, everything sounds ultraloud instead of being ultrareduced? shouldn't 9db then drop to -19 (i'm a poor mathemathic)?

has anybody an easy to understand answer?

thanks for your attention
 
No, you are misunderstanding one part. The ratio of 3:1 means that, (once above the threshold), for 3db of increase at the input you only get 1 db of increase at the output. Thus, 12db plus the 1db increase equals 13db at the output...Instead of 15db like the source wanted to get away with.

(I am speaking in terms of basic Desibel scale where an increase in volume means more db.)

Happy compressing...
 
Last edited:
Thus, 12db plus the 1db increase equals 13db at the output...Instead of 15db like the source wanted to get away with.

urm... then, in this case, would this be the moment when the gain reduction meter thing reads -2db?

so if the threshold where -40db and a 3:1 ratio... to a, lets say, -15db of input gain, we should have to add... uh... about 6 or 7 dbs to the initial -15 db, resulting in -8 db of output?

i wish my hobby were planting potatoes...

thank a lot for your help, rawdepth
 
I always wondered this, too. If compression is all about gain *reduction*, then why do compressed signals always sound *louder*?
 
Gain reduction is the means, not the end. The purpose of a compressor is to restrict dynamic range. If you are looking just for gain reduction, turn the gain down. Smaller dynamic range can increase the apparent loudness of a track compared to the uncompressed version of the same track.

When the dynamic range is restricted, signals of lower level are closer in volume to signals of higher level than they were before, giving a track more consistent loudness from beginning to end. So every low-level noise like breath noise, for instance, is louder when you use make-up gain to bring the track volume up to where you want it.
 
Yes, what he said.

Or in even simpler terms...

With compression, if you lower only the highest volume peaks, but let the rest unchanged, then you have the luxury of raising the overall signal level without any volume peaks clipping, or "hitting the ceiling" so-to-speak.

With the peaks leveled off a bit, you simply turn up the output gain and PRESTO...louder music.

For instruments like drums, compression is used to keep volume spikes under control. For example, if the average number of drum hits is set to a good recording level, there may still be a stray drum hit once in a while which pegs the meter. Compression can fix that without having to turn down the overall recording level (or playback/mix level) just to accommodate that one loud hit.
 
If you are using a compressor with auto makeup gain, the plugin automatically turns the gain up to make up the gain that it has taken away.
 
so compression reduces the dynamic range of a song... let's see...

i've recorded about a dozen songs, some rather metal, some rather electronica, some pretty loud parts, some pretty soft parts... not a dozen punk rock songs repeating the same strum from start to end.

then...

why do they sound so f***ing lineal when you listen to the entire playlist? it is as if the word "crescendo" were not in my dictonary.is this problem related to compression? they all have a final stage of multiband compression (masterx3, with very similar parameters from one song to another...hmm...) and some eq after it, but the compression ratio is not that much (sometimes about 2:5, sometimes about 1:8) and none of the instruments is specially overcompressed at the recording/mixing stage.should I manually pump up the gain of louder parts, or reduce the softer ones? should i automate the final compression plugin? adding another multiband plugin? any whatever?

maybe i should post it in another thread... dunno

and thanks to everyone for your feedback!
 
this belongs in Mixing/Mastering.....

but why not continue here and wait for it to be moved.

verytorpe said:
so compression reduces the dynamic range of a song... let's see...

i've recorded about a dozen songs, some rather metal, some rather electronica, some pretty loud parts, some pretty soft parts... not a dozen punk rock songs repeating the same strum from start to end.

then...

why do they sound so f***ing lineal when you listen to the entire playlist? it is as if the word "crescendo" were not in my dictonary.is this problem related to compression? they all have a final stage of multiband compression (masterx3, with very similar parameters from one song to another...hmm...) and some eq after it, but the compression ratio is not that much (sometimes about 2:5, sometimes about 1:8) and none of the instruments is specially overcompressed at the recording/mixing stage.should I manually pump up the gain of louder parts, or reduce the softer ones? should i automate the final compression plugin? adding another multiband plugin? any whatever?

maybe i should post it in another thread... dunno

and thanks to everyone for your feedback!


Your compression problems sound farmiliar, I think everyone who is trying to grasp the concept (including me) for the first time has many frustrating moments. Don't worry it will all make sense as long as you don't toss that MF into a brick wall or something.

You've used compression (and multiband compression) on all of your songs, most with the same settings, and you say your album is very level linear (or lineal?).

First, take off all the compression (if possible) you've put on. Proceed through the mastering process. While comparing song levels (with your ewars of course, not meters), if a song in the playlist needs to be louder, only then would you bring some compression into the picture. If you need it to be 3db louder, set the threshold to just above the average level (RMS) of the song, and adjust ratio until you have a 3db reduction, then add the 3db back with output gain.

If you don't NEED compression, don't use it. If you have one peak that keeps you from turning up, digitally edit it out by zooming in close, selecting the peak, and reduce the level manually. I've done this "manual" compression for entire songs, takes a lot of patience, but keeps you from getting too compressed. Why compress the entire song when you can manually remove the few problem peaks? This is just something to think about.

To avoid compression I also "ride the envolopes". If you have multitrack software, try using the volume evelopes to create crescendos and climax moments of a song.

My main point is not to depend to much on compression right now, while your still trying to figure it out. Use you imagination and common sense to come up with solution for YOUR own unique problems. If compression sounds bad to you, don't use it.

By the way, your ratios of 2:5 and 1:8 don't make much sense to me.

Ratios are the following:

input gain : output gain

i.e. 2:1

For every 2 db's over the threshold, 1 db come out of the compressor.

hope this helps,
Later.
 
RawDepth said:
With compression, if you lower only the highest volume peaks, but let the rest unchanged, then you have the luxury of raising the overall signal level without any volume peaks clipping, or "hitting the ceiling" so-to-speak.

Actually, what you are describing here (strictly speaking) is limiting, not compression.

Compression squeezes everything above the threshold. So the loud becomes quieter and the quiet becomes louder--above the threshold. It's not just the peaks that get processed, like when limiting. Anything below the threshold is not affected.

The reason why the track seems louder is that you are probably raising the output gain. So since the louds are softer, you bring those up. And since the softs are louder, those get brought up as well. So the perceived volume of the track is louder, since the average volume of the track is louder. Also, since you set the threshold at -40, you are probably getting most of the dynamic range of the track involved in the compression process. That's a very different sound than if you set the threshold at -8 (for example), where you would be compressing a smaller portion of the dynamic range of the track. Basically skimming the top at that point.
 
with "lineal" I meant "linear"... my english is worse than my spanish and my spanish is even worse than my typing abilities.

actually, the -40 db threshold was just and example... but thinking about it, i've probably used, if not -40, maybe -38 or -35 db theshold in some parts... i don't remember; and yes, everything went just louder, I couldn't appreciate too much benefit.

the 2:5 or 1:8 thing... who knows, that's what reads on the "ratio" display of the masterx3 compressor. I simply use to move the slider up and down without caring too much about what it reads, I try to use my ears (not very succesfully, anyways)

and well... that "take off every compressor" approach seems to be a good starting point. I'm moving the previous message to the mixing/mastering thread, but I would like to make a last question first:

In the same way I have compressors everywhere, there was a time in which I had eqs everywhere. somewhere I read that such amount of eq was unnecesary, a similar way to difference one instrument from another was the use of compression... so I got rid almost all the eqs I had and started to add compressors here and there. Through this process, my ears have learned somewhat to add eq more carefully; now it's time to teach them about compression...

this question is not specially related to the "boring linear audition" question, but I would like to know if any of you have something to say about it


moving to the mix/master forum. thanks again!!! see ya...
 
SonicAlbert said:
Actually, what you are describing here (strictly speaking) is limiting, not compression.

Compression squeezes everything above the threshold. So the loud becomes quieter and the quiet becomes louder--above the threshold. It's not just the peaks that get processed, like when limiting. Anything below the threshold is not affected.


No, I disagree.

Compression does not raise the level of any signal. The output preamp can raise the overall volume after the compression stage but, the compression circuitry only lowers volume.

The difference between compression and limiting is as follows:

With compression, all signals above the threshold gets reduced, but only by an amount controlled by the ratio. If ratio is 4:1 then the signal gets reduced by 75% of the total amount over the threshold. (if signal is 4db above threshold, then it is reduced until it is 1db over threshold.)

With limiting, all signal above the threshold is reduced until it is equal to the threshold. (assuming that attack and release are set properly to capture it all.)

Everything below the threshold is unchanged in both cases.

A compressor can be made to act like a limiter by turning the ratio all the way up to 100%.
 
Last edited:
Back
Top