Histograms, ranges and ratios are only a small part of what makes Dynamic Grading tick. It’s the tiny details that turn it into the dynamics powerhouse that it is. One of those details is how the passage of time is handled.
One of the driving thoughts behind the development was to create a really versatile yet simple to handle dynamic processing engine that – just for once – does it right. And when you hear me saying that it should be done right, that implies that most of the time it’s done wrong, right?
Now recall the core purpose of any dynamics processor: measure the variations of short-term loudness (= dynamics) and change them in a meaningful way.
Of course there’s the thing about properly dealing with the spectral decay of real-world audio signals, which makes a huge difference already. But there’s another tricky bit involved in the about the above declaration of intent. And again, most conventional compressors get this wrong.
For the purpose of loudness, momentary amplitude or power is almost meaningless. The perception of loudness is instead based on averaging power over a short period of time. So any dynamics processor must do such time averaging in order to do its job. Without it, a compressor for example becomes a saturator.
The Passage of Time
Averaging over time obviously requires time to pass. Obviously, we can detect events like transients or onsets only after they have happened. However, in the moment they happen, the average is still dominated by whatever happened before. It takes some time for the average to fully reflect the fact that the snare drum has just been hit.
Averaging over time is equivalent to low pass filtering, and unless the filter can look ahead into the future, it will delay the result.
What conventional compressors or expanders do is exactly that, they average their level measurement over some time with a lowpass filter, typically controlled by the attack parameter, then apply a gain based on the result.
But here’s the problem: the gain – based on the averaged level measurement – is delayed compared to the audio it’s applied to. Consequently, the conventional dynamics processor is always too late. The peak that a compressor is supposed to compress is already gone when it finally gets to it.
Looking Ahead
Now the solution is obvious. We delay the audio just as much as the delay from the averaging, and everything’s aligned again. And that’s what Dynamic Grading simply does without making a big fuss about it. It just works. For some reason, even compressors that do have a lookahead feature will have it switched off by default and make you worry about finding the correct setting.
But of course, most compressors don’t have a lookahead anyway, because for some reason it’s more important to carefully model how the actual attack time changes when you change the ratio in some piece of equipment built in the 70s. Because why should you get the 10ms attack time you requested by setting that knob to 10ms? Ah, I’m going on one of those tangents again.
Now you might say that being too late is actually good, because by increasing the attack on a conventional compressor, you can get snappier transients. That is kind of true, but as Dynamic Grading users we simply do this by adjusting the punch range, which gives us much better control about how much snappiness we want, and we don’t have to trade it off against the dynamic range compression in the body range that we mainly want to achieve.
That said, if you long for the particular frustration of getting a conventional compressor to give you just the right combination of compression, punch and timing by fiddling with the attack time, Dynamic Grading’s got you covered as well. Just set the latency mode to “Live”, which removes the lookahead and makes the response time slider act like the attack time of a conventional compressor.
If you haven’t tried out Dynamic Grading yet, this article might have whetted your appetite. Don’t hesitate to check out the free trial, and don’t forget to reach out to us if you have any comments, questions or suggestions!
