Quote:
Originally Posted by Nos402
So is there really any reason to use a compressor on the audio as many audiobook narrators/engineers do? Isn't this accomplishing the same thing?
|
ashcat_lt is exactly right, both that RMS normalization and compression do different things and also that even if other narrators need compression and limiting it doesn't mean that your performance does also. I do a lot of narrative audio, both sending off unmastered files and mastered files, and I would add that if it sounds good and they've been accepting it without any compression that's terrific. But it would be good to experiment with a small amount of it to see if you find it improves it even more, especially since, yeah, there's pretty much always a compressor and limiter in the chain on a narrator : )
Of course it totally depends on the material. Fiction with characters making loud exclamations every once in a while is much different than a non fiction talking about Health Care. The former has a wider dynamic range while the latter, since it's not varying from a narrative tone, has a narrower dynamic range. The this, which is sort of what I usually have in there: have the compressor as suggested by ashcat_lt followed by a limiter, which could even be the same ReaComp set as a limiter, and don't have either set so that either is reducing by much by themselves. Have the first ReaComp with a low ratio (2:1 or 3:1) with a moderate attack and slowish release. Then have the second set at a higher ratio, 10:1 or 20:1, and a faster attack and release. Not super fast but say 20mill attack and 100 release. See if you can set their thresholds so that both are just licking the transients down by, say, 2 db usually, 3 db tops. One fancy pants difference you can add is to have a soft knee on the first and a hard knee on the second, as that's more limiter-like.
The idea is that the first will do it gently and barely, so that transients will get through, and then the second will knock down those transients but won't kick in except for those peaky transients. Then use the RMS Normalizer to put the RMS level exactly where you want it. It works better, or at least more subtly, than having a single dynamics processor take the peaks down 4 db, if that was the goal.
If it's a clean recording and the room ambience is negligible gating is probably not necessary. But bringing the peaks down by 4db has the same noise floor as the noise floor being raised by 4 db, so adding a gate so that it only lowers down 4 db less (never completely to silence,which is what many people think of when they hear the word "gate") and only to drop when you're not speaking at all, is another common narration tool to get the noise floor back where it was before processing. You won't hear at all if it's very subtle. Same as the compressor and limiter. You can make them so you don't know they're in at all. Or maybe you like the sound of one of them hitting a little harder than that. If a narrator has a wide dynamic range and theres a lot of whispering and screaming I'll have both hitting pretty hard at the loudest in order to keep it together (or even add another). If the program does not have that aspect, and it sounds like yours doesn't, you could still use the same chain just have the thresholds set more conservatively.
Reaper is awesome here because the included plugins are truly good enough to use. And with offline rendering you can do a small patch and listen back on another device very quickly and easily.
Have fun!
RMS Normalization and Compression
Linear Mode
