How to hear the compression

[apaclin]

Also a question. I noticed that 1176 (using Arturia's one, but same story with IK Multimedia) kills the transient way more on high ratios, than on 4:1 (both with the slowest attack). Why is that?

[Drew Stephenson]

If I'm understanding the question right, because if the time taken to get to full attenuation is the same, then before it's got to that full attenuation state the higher ratios will still be having a stronger effect than the lower ones.
If you've got two ratios, say 4:1 and 8:1, and they're both set to come in after 10ms, then at 5ms your higher ratio will probably still be taking twice as much off the transient even if it's only halfway to full attenuation. That won't make much of a difference to sounds with a slower attack but will affect fast transients.

[apaclin]

If I'm understanding the question right, because if the time taken to get to full attenuation is the same, then before it's got to that full attenuation state the higher ratios will still be having a stronger effect than the lower ones.
If you've got two ratios, say 4:1 and 8:1, and they're both set to come in after 10ms, then at 5ms your higher ratio will probably still be taking twice as much off the transient even if it's only halfway to full attenuation. That won't make much of a difference to sounds with a slower attack but will affect fast transients.

But 1176 doesn't just change the shape of the transient, it makes it quieter with high ratios, which is not logical. The transient will be quieter after 5ms with the higher ratio than with the lower one, but at 0ms they should be the same.

[Wonks]

You need to go back and re-read articles on how compression works.

For a start, we are dealing with an attack between 0.02ms and 0.8ms, so even at its slowest attack, an 1176 will severely limit even fast initial transients.

The compression kicks in after the signal exceeds the threshold. The 1176 uses a peak detector, so the moment the signal crosses the threshold, compression/limiting kicks in. If the ratio is 4:1, then for every 4dB increase in input signal level, the output signal only increases by 1dB.

If the ratio is 8:1, then for every 8dB increase in input signal level, the output signal only increases by 1dB. Or for a 4dB increase in input, you only get a 0.5dB increase in output.

So the higher the ratio, the more compressed (and quieter) the output signal is compared to the input signal.

That’s before any make-up gain is considered. More compression = less dynamics in the signal. The loudest peaks aren’t as far above the quietest sounds as before.

You don’t need to use make-up gain, but if you do, and adjust the gain so the output peaks are at the same level as the input peaks, then the average energy of the signal has been significantly increased and the resulting sound is louder.

A faster attack will let less of the transient through, so for the same ratio setting, a faster attack will result in a more ‘squashed’ waveform. As its peak signal level will be less than with a slower attack setting, you need to add a bit more make-up gain to get the output peak to match the input peak, so the overall signal will be a little bit louder.

But the faster the attack, and/or the higher the compression ratio, the more the waveform is distorted/squashed away from the original, the dynamics are reduced, and you can easily turn an interesting sound into a dull (but loud) one.

[apaclin]

You need to go back and re-read articles on how compression works.

For a start, we are dealing with an attack between 0.02ms and 0.8ms, so even at its slowest attack, an 1176 will severely limit even fast initial transients.

The compression kicks in after the signal exceeds the threshold. The 1176 uses a peak detector, so the moment the signal crosses the threshold, compression/limiting kicks in. If the ratio is 4:1, then for every 4dB increase in input signal level, the output signal only increases by 1dB.

If the ratio is 8:1, then for every 8dB increase in input signal level, the output signal only increases by 1dB. Or for a 4dB increase in input, you only get a 0.5dB increase in output.

So the higher the ratio, the more compressed (and quieter) the output signal is compared to the input signal.

That’s before any make-up gain is considered. More compression = less dynamics in the signal. The loudest peaks aren’t as far above the quietest sounds as before.

You don’t need to use make-up gain, but if you do, and adjust the gain so the output peaks are at the same level as the input peaks, then the average energy of the signal has been significantly increased and the resulting sound is louder.

A faster attack will let less of the transient through, so for the same ratio setting, a faster attack will result in a more ‘squashed’ waveform. As its peak signal level will be less than with a slower attack setting, you need to add a bit more make-up gain to get the output peak to match the input peak, so the overall signal will be a little bit louder.

But the faster the attack, and/or the higher the compression ratio, the more the waveform is distorted/squashed away from the original, the dynamics are reduced, and you can easily turn an interesting sound into a dull (but loud) one.

Apparently you haven't read what I just wrote

[Wonks]

I don't think I've fully understood what you wrote even after reading it several times.

The 1176 has a soft-knee characteristic, and the 'fixed' (non-adjustable) threshold value actually varies with the compression ratio. The graph below shows the measurements from an original hardware 1176.



(It looks like the results were achieved using a fixed amount of make-up gain of around +20dB).

You can see that as the ratio increases, the point at which compression starts to affect the signal also increases and the 'knee' becomes harder. I wasn't aware of this until just now, but it was certainly a feature of the first Urei 1176 units. Later versions may have slightly different characteristics.

So the 1176 is far from a textbook compressor, where the threshold stays as set as the ratio is increased.

We then come to how your particular software emulation of the 1176 has been implemented. The software designer may have decided to alter the knee characteristics or cut-in points so that start at the same point. Or they may have added a look-ahead function which looks further ahead as the ratio is increased, so that it performs better as a limiter at stopping big transients.

One 1176 emulation could act quite differently to another one depending on how it's been coded or how the hardware model they were using differed from other models or how it was internally calibrated).

So without doing a full test on the software 1176 you have, using a tone generator and increasing the output in 0.1dB steps and mapping the output level for different compression ratios, you just won't know exactly how the compressor behaves. In theory (according to that chart), a 4:1 setting should certainly cut in at a lower threshold than at 8:1, so a signal that just starts compression at 4:1 shouldn't cause any compression at an 8:1 setting.

So, like a lot of answers on this forum ' it depends'.

[Hugh Robjohns]

I noticed that 1176 (using Arturia's one, but same story with IK Multimedia) kills the transient way more on high ratios, than on 4:1 (both with the slowest attack). Why is that?

Fundamentally, it's because a higher ratio imposes much more gain reduction when the signal exceeds the threshold. The signal is squashed much more aggressively with a high ratio setting compared to a low ratio (for a given threshold level).

The reference to there being a level difference at 0ms (before the compression has had time to evolve) is, I suspect, related to automatic make-up gain compensation.

It's also worth bearing in mind, when using modelled vintage plugins, that these are inherently non-linear systems based on imperfect analogue designs which often suffered control interactions. The numbers on the control scales are rarely accurate, or directly comparable with other devices, to start with, and things like attack and release times often vary with different ratio values anyway because of inherent circuit interactions.

At the end of the day, all that matters is whether you can achieve the sound you want by twiddling the knobs. An understanding of the process helps in twiddling the knobs the right way and by the right amount... but every compressor has different characteristics due to its unique design — they are never directly comparable regardless of the panel control setting scales. Thats why there are many different compressors in use.

[tea for two]

Twiddling is well recommended lol.

Where EQ, Filtering, Transient shaping isn't quite getting the sound we are after can use Compressor to shape a sound.
For instance software emulation dbx160 (bundled with Logic) I've used can shape a Bass, Heavy Guitar to be aggressive to cut through a meaty mix.