Compression: Ratio higher, compression higher. Gain reduction... less??

Jonny, can you post the link to the article? I would like to read it to see exactly the math the author is talking about.

Thanks

Here are the quotes:

"Compression – or gain reduction, if you will – is measured in terms of a ratio of input volume to output volume. For example, a compression ratio of 2:1 (pronounced “two to one”) means that for every 2dB a signal rises above the threshold going into the compressor, only 1dB above the threshold will come out. Another way to think of it is to take the number of dBs a signal rises over the threshold on the input and divide that amount by the first figure in the compression ratio setting to get the number of dBs that signal will rise above the threshold after compression.

Let’s say that we have an incoming signal that peaks at 0dB. We have our threshold set to –12dB and our ratio set to 3:1. This means that the incoming signal peaks at 12dB over the set threshold. For every 3 of those 12dB going in, 1dB will come out, meaning that after the 3:1 compression, the signal will now peak at 4dB (12 divided by 3 equals 4) above the threshold, which means that our 0dB peak will now instead peak at –8dB."

(And further on he kinda returns to it...)

In the example we used to describe how gain reduction and compression ratios worked, we used an example that showed that a 0dB peak that went through a compressor set to 3:1 compression at a threshold of –12dB would come out compressed to a peak level of only –8dB, an 8dB reduction in gain.

Take that same example from the above paragraph. Now change the threshold setting from -12dB to –6dB. Now that same 0dB signal is only 6dB above the threshold value. So even though the compression ratio remains the same at 3:1, the compressor now takes those 6dB and divides them by 3 for an output gain of 2dB above threshold. This now puts the peak of the compressed signal at –4dB instead of –8dB.

And then further on again he just has me in a mental tangle, where he takes a statement by an audio engineer and shows why when someone says "I compress at 4:1" it means nothing without an idea of the threshold or the dynamic range etc...

“Vocal: a good starting place for a lead vocal is 4:1 ratio, medium attack and release, and a threshold set for about 4 to 6 dB of gain reduction.”

The engineer here is assuming that there actually is 4 to 6dB of reduction to be had from the vocal signal, and that 4 to 6 dB is the right amount for that particular vocal signal. He is making assumptions on the levels and dynamic range of the vocal signal. There can be much more or less, depending on the quality of the tracking.

The properties of an average input signal from most of us can vary greatly from those that the pro mixing engineer is used to dealing with. These properties have an important effect on the math as well. Let’s go back to the original compressor settings of 3:1 compression with a threshold of –12dB. But let’s say that the input signal peaks at –3dB instead of 0dB. Now the peak rises only 9dB above the threshold. After compression, the output signal will peak at –9dB and not the –8dB of the original example. Now, 1dB difference may not seem like much, but look at it in terms of the difference in overall dynamic change. In the original example, the peak was compressed from 0dB to –8dB, an 8dB difference. In the second example the peak was compressed from –3dB to –9dB, a 6dB difference. There is a difference of 2dB in the amount of overall dynamic compression between the two. This 2dB can be very significant in the average (RMS) volume of the signal – especially when you start summing tracks together – and therefore they make just as significant of a difference in the perceived loudness of the recording.

Add the two variables of threshold and signal strength together and in this case we could have a compression result that can vary by a total of 6dB or more for a single, common compression ratio. In the digital realm, where we’d be talking digital 6dB, that’s a doubling or halving of your volume, depending on whether you’re going 6dB up or 6dB down.

This is why the engineer in that quote makes a point of saying that one must adjust the settings to achieve a certain amount of gain reduction, not that the settings are rigid. In this case they say to adjust the threshold, but in reality one could keep the threshold setting rigid and vary the compression ratio instead, and still wind up with the same amount of reduction. In fact, adjusting ratio instead of threshold is preferred when your threshold gets dangerously close to the part of the signal you don’t want to compress.

Yep, you got it. It's just hasn't quite clicked.

He covers it here:

In the example we used to describe how gain reduction and compression ratios worked, we used an example that showed that a 0dB peak that went through a compressor set to 3:1 compression at a threshold of –12dB would come out compressed to a peak level of only –8dB, an 8dB reduction in gain.

The signal at 0db is 12db over the threshold, divide by 3 and you get 4db over the threshold.The signal out is -8db (-12db plus 4). That's a gain reduction of 8db.

Brett

So I guess if you knew what sort of dynamic range you wanted, all you need to do is set your threshold and 'backwards calc' to get your ratio (or a starting point anyway)

Depends. Most people will go for "ballpark" settings for ratio, attack and release (from experience) and then adjust the threshold until they'll see (or better hear) the desired gain reduction. That's why the gain reduction meter is so important on a compressor. Then tweak attack/release/knee, rinse and repeat until it's right and finally adjust make-up gain (if at all). At least that's how I'm doing it...

werner