Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Hard limiting vs. single-band compressor (Read 4050 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Hard limiting vs. single-band compressor

This is probably a simple question for those who are really into audio editing, but since I'm just an enthusiast, and definitely not educated in an audio-engineering way, I need some help.

So I'm making a medley of music, and in the mixdown, the originally 0 dB peaks are at -3 dB (to give room for later work, I get that of course), but the parts where 2 loud tracks get mixed, it goes up right to 0 dB, often. However, I find that lowering volume in multitrack editing changes the intended experience, so I abandoned that idea. Correct me if I'm wrong, but I see the solution in either hard limiting the mixdown, or using compression, so the original peaks can be close to 0 dB once again (-0,1 or -0,2 or so). But which one?

The source music is already mixed and mastered, so it can only be single-band compression, I know that much (at least I think I know that right), so what's better for this purpose, hard limiting or a single-band compressor, and if the latter, have you got any suggestions for settings? Thanks so much

Hard limiting vs. single-band compressor

Reply #1 –
If you want the peaks to be really close to 0 dbFS then you are better off with a look ahead (hard) limiter than a regular compressor. A regular compressor usually don't have as fast attack as the first mentioned one do you can not be sure that it will absolutely stop all peaks from going over 0 dBFS.

Hard limiting vs. single-band compressor

Reply #2 –
Quote from: punkrockdude on
If you want the peaks to be really close to 0 dbFS then you are better off with a look ahead (hard) limiter than a regular compressor. A regular compressor usually don't have as fast attack as the first mentioned one do you can not be sure that it will absolutely stop all peaks from going over 0 dBFS.


About what you're saying (if I got it right), the compressor doesn't have to do the gain, it's perfectly fine if it just "flattens" the new peaks above -3 dBFS to make the whole thing more equal, considering max peaks (btw I'm using 16 bit sources and editing in 32 bit, so I guess there's enough room for leaving some safety space for the compressor and then normalizing).

My point is, which one is less destructive to the sound, and also (since I'm not a pro), is there a way to set the compressor up to only deal with the new, big peaks, and leave everything below -3 dBFS alone, so to speak. Because if it cannot work that way, I have no use for it really.

Hard limiting vs. single-band compressor

Reply #3 –
You need to understand what the terms mean first. 

A compressor, especially as it exists in software today, is highly flexible.  You have control over the threshold (point above which it begins to reduce gain), compression ratio (the ratio of input change to output change, also known as slope), the attack time, release time, and detector type (peak, average or RMS).  With all that available, you can't just call it a "single band compressor", too many variables.  For example, set it to a high ratio like 30:1, or infinite, set the threshold at -3dBFS, set the detector to Peak, and set the attack time to less than 1ms, you've essentially got a peak limiter.  But, set it for a 2:1 ratio, 50ms attack, 1 second release, -20dBFS threshold, and RMS detector, you've got a really gentle slow compressor. 

A "hard limiter" usually means it's a clipper: a gadget that just lops off peaks above a threshold.  These things are useful tools, but not by themselves.  Clipping off a peak, if it's short and very occasional is usually inaudible, but slamming into it regularly causes very high intermodulation distortion, harmonic distortion, and sound just plain nasty.  A hard limiter should be used with great care and caution, and usually preceded by a regular limiter of some sort, perhaps a high-ratio peak responding compressor.

A "peak limiter" or just "limiter" is a very fast attack, infinite ratio, high threshold, peak responding gain controller.  "Look Ahead" is essentially a negative attack time, responding with gain reduction to a peak it hasn't actually passed yet. Nice trick!  However, the release time is also a distortion control: the faster the release, the more distortion.  And the harder you hit it, the more distortion will result.  Some of that is "effect" and desirable some times.  But it's distortion regardless.

For your application it sounds like you need to use a peak limiter, not a "hard limiter", fast attack or look-ahead a few ms, and start with about a 50ms release time.  Watch it's gain meter, don't allow more than about 3-5dB of gain reduction on those peaks you're trying to control, and listen to the result.

Don't worry too much about 16 bit source files, you're working in 32bit float, so no problem.  Remember, linear quantization (that's what we all use) has more resolution at the loud end than the quiet end, so peak processing works pretty well.  If that doesn't make sense, somebody can ask, I'll clarify.  Where you need more headroom to work in is for functions like mixing and EQ, where you have the possibility for large increases in peak level that could exceed full-scale, sort of the opposite of what happens with a peak limiter.

For the sake of completeness, and at the expense of confusion, multi-band processors exist too.  They split the spectrum into bands, then compress and limit within those bands (usually without regard for the other bands!), then sum it all back together.  Where sounds like a loud kick drum would modulate the entire mix using a single-band processor, a multi-band processor would work on the kick drum by itself without modulating the rest of the mix...at least, not too much.  Multi-band lets you process more without side effects, true, but it also ends up being a dynamic equalizer, like someone constantly fiddling with a 5 band graphic.  It can work, but these are actually fairly difficult to set up. Sadly, radio stations use them to process all their on-air audio, so if your mix is going to be on radio, you don't want to do what the station will do anyway.  Just makes things worse. 

Ok, more than you asked for, but you probably needed it...
 

Hard limiting vs. single-band compressor

Reply #4 –
Thanks for that very useful and detailed answer, greatly appreciated

Just one tiny thing:
Quote from: dc2bluelight on
With all that available, you can't just call it a "single band compressor", too many variables.

I'm not calling it one, Adobe people do (in Audition). There's a single-band compressor (well, 2, because there's an iZotope one as well), and a multiband compressor, too (which you talked about later, and I think we agree it's probably not the thing to use, since the music has been fully treated and even though this isn't going into any kind of production - actually a medley of Tomb Raider game music, so most listeners will be game fans - it's still important to keep the the original sound wherever possible.

You're definitely right about needing a peak limiter, since what I want is no change whatsoever to the parts of audio that don't go above -3 bBFS in the mixdown (which I'm not sure, or rather don't know, if a compressor is capable of).