Limiting Stems & Submixes

[minilog]

I guess the idea is to cut down on peaks and smoothen things out so the final master limiter doesn't work so hard. Interesting concept that I never heard of until today. Has anybody done this who can share some tips? Which limiter do you think would be good for this? What amount of gain reduction would you be going for when limiting a submix?

[Zac]

I guess the idea is to cut down on peaks and smoothen things out so the final master limiter doesn't work so hard. Interesting concept that I never heard of until today. Has anybody done this who can share some tips? Which limiter do you think would be good for this? What amount of gain reduction would you be going for when limiting a submix?

I'd be interested to hear some tips on this too. One of the reasons I've never done it in a finished song is that I've worried that an already limited bus might hit the final limiter and sound bad.

[Loque]

I always try to eliminate the peaks ASAP in the mix. I use everything that might be suitable, from EQ, compressor, limiter, dynamic EQ, multiband stuff, saturation, transient shaper, sound adjustments,... Best thing is a true peak limiter with adjustable knee, but they introduce latency AFAIK. Still need to get more experience here...

[minilog]

I always try to eliminate the peaks ASAP in the mix. I use everything that might be suitable, from EQ, compressor, limiter, dynamic EQ, multiband stuff, saturation, transient shaper, sound adjustments,... Best thing is a true peak limiter with adjustable knee, but they introduce latency AFAIK. Still need to get more experience here...

I've never used a limiter in the mix for anything! What would make somebody reach for a true peak limiter as opposed to the other options mentioned? I'm really big on transient shaping though. I almost bought another transient shaper yesterday actually even though I already have DCAM and elysia Nvelope.

Something that "peaked" my interest was the Brainworx bx_limiter which they don't claim outright to be a mastering limiter but rather suited for stems and submixes.

[Loque]

I always try to eliminate the peaks ASAP in the mix. I use everything that might be suitable, from EQ, compressor, limiter, dynamic EQ, multiband stuff, saturation, transient shaper, sound adjustments,... Best thing is a true peak limiter with adjustable knee, but they introduce latency AFAIK. Still need to get more experience here...

I've never used a limiter in the mix for anything! What would make somebody reach for a true peak limiter as opposed to the other options mentioned? I'm really big on transient shaping though. I almost bought another transient shaper yesterday actually even though I already have DCAM and elysia Nvelope.

Something that "peaked" my interest was the Brainworx bx_limiter which they don't claim outright to be a mastering limiter but rather suited for stems and submixes.

I am not the mastering guru, but it has something to with digital->analog conversion, where signals may clip again, even if they have digitally a perfect limited value. I found this, that explains it quite good:
https://www.masteringthemix.com/blogs/l ... k-metering

You can checkout Yolean Lodness meter that displays True Peak and its for free.

Here are a few limiter, but dont have much experience with them yet:
https://theproaudiofiles.com/free-mastering-plugins/

[selig]

Using limiters and compressors on sub-mixes/stems is pretty common, as much as any approach is "common". There will be those who say "never" and those who say "always", and then 90% of the rest fall between those extremes!

IF going this route, best to start subtle, especially if hard limiting. I'm a fan of using smaller amounts of compression at many stages, but tend not to need outright limiting when using this approach (saving it for the final mastering stage). It's easy to over do it, for sure.

In my experience, limiting works best when the signal is already well controlled, to deal with those few peaks that keep the signal from being as loud as you want it to be - or those few peaks which would clip, in cases where loudness is not at all a part of the equation.

Meaning, a very dynamic signal won't be helped much by limiting since limiting will only affect the very loudest parts. That's why the very general "rule" is to use gentle ratios/compression at the earlier stages, increasing as you get to the final stage. In traditional recording scenarios this would mean recording basic tracks with gentle compression (playing it safe at this stage, since you can't easily "undo" recorded compression) and taking advantage of some tape saturation. Then using more aggressive compression during mixing, and finally using limiting for mastering. Again, a VERY general rule…

For modern productions ITB with software like Reason, this would equate to using gentle compression on individual channels, using more aggressive compression on busses, and using limiting on the final master.

One exception is for parallel compression, where a limited parallel channel (when mixed in equal parts with the original/dry channel) equates to a ratio of 2:1 compression. I often add saturation when going this route, because I prefer parallel saturation/limiting to the serial version in many cases (especially when trying to be more subtle).

[RobC]

Using limiters and compressors on sub-mixes/stems is pretty common, as much as any approach is "common". There will be those who say "never" and those who say "always", and then 90% of the rest fall between those extremes!

IF going this route, best to start subtle, especially if hard limiting. I'm a fan of using smaller amounts of compression at many stages, but tend not to need outright limiting when using this approach (saving it for the final mastering stage). It's easy to over do it, for sure.

In my experience, limiting works best when the signal is already well controlled, to deal with those few peaks that keep the signal from being as loud as you want it to be - or those few peaks which would clip, in cases where loudness is not at all a part of the equation.

Meaning, a very dynamic signal won't be helped much by limiting since limiting will only affect the very loudest parts. That's why the very general "rule" is to use gentle ratios/compression at the earlier stages, increasing as you get to the final stage. In traditional recording scenarios this would mean recording basic tracks with gentle compression (playing it safe at this stage, since you can't easily "undo" recorded compression) and taking advantage of some tape saturation. Then using more aggressive compression during mixing, and finally using limiting for mastering. Again, a VERY general rule…

For modern productions ITB with software like Reason, this would equate to using gentle compression on individual channels, using more aggressive compression on busses, and using limiting on the final master.

One exception is for parallel compression, where a limited parallel channel (when mixed in equal parts with the original/dry channel) equates to a ratio of 2:1 compression. I often add saturation when going this route, because I prefer parallel saturation/limiting to the serial version in many cases (especially when trying to be more subtle).

If I remember correctly, Normen suggested something like - to my memory/understanding ~ that if there's a great mix, it shouldn't be limited/compressed.
I guess as long as the mix falls between -23 and -16 LUFS, we're mostly good. Later, a record label, maybe, will make it -6 LUFS anyway for no reason. xD
Other than that, my thought was multi-band saturation; or just clipping. Sometimes a transient, in my experience, sounded better with hard clipping, than soft clipping with rounded edges. It's just there's that thought in my mind, where people say, 'don't clip!' because of ear fatiguing. Could be I take some things too literally, as when throughout a song, there are say, 5-10 hard clips, that are a few sample bits long; and there's a loudness war master that constantly clips ~ well, it's not the same.

[minilog]

One exception is for parallel compression, where a limited parallel channel (when mixed in equal parts with the original/dry channel) equates to a ratio of 2:1 compression. I often add saturation when going this route, because I prefer parallel saturation/limiting to the serial version in many cases (especially when trying to be more subtle).

Parallel processing seems like an interesting approach. I actually do very little parallel compression and usually reserve that technique for saturation or to collapse a channel to mono for compatibility.

[Zac]

One exception is for parallel compression, where a limited parallel channel (when mixed in equal parts with the original/dry channel) equates to a ratio of 2:1 compression. I often add saturation when going this route, because I prefer parallel saturation/limiting to the serial version in many cases (especially when trying to be more subtle).

Parallel processing seems like an interesting approach. I actually do very little parallel compression and usually reserve that technique for saturation or to collapse a channel to mono for compatibility.

I actually had softube's dynamite limiting a synth today (quick fix) and was struggling to get the effect amount right and thought about putting it in parallel to make it easier to control the effect amount but then thought the latency would be an issue... solved it by adjusting the output level.

P.S. Wet/dry knobs are always welcome devs... !

[minilog]

I actually had softube's dynamite limiting a synth today (quick fix) and was struggling to get the effect amount right and thought about putting it in parallel to make it easier to control the effect amount but then thought the latency would be an issue... solved it by adjusting the output level.

P.S. Wet/dry knobs are always welcome devs... !

Thanks for bringing the Dynamite to mind. I totally forgot about that one and haven't used my trial yet *grins like a grinch who stole dynamics*.

[Zac]

I actually had softube's dynamite limiting a synth today (quick fix) and was struggling to get the effect amount right and thought about putting it in parallel to make it easier to control the effect amount but then thought the latency would be an issue... solved it by adjusting the output level.

P.S. Wet/dry knobs are always welcome devs... !

Thanks for bringing the Dynamite to mind. I totally forgot about that one and haven't used my trial yet *grins like a grinch who stole dynamics*.

I really haven't found it to be what the Softube literature claims until now. In fact after going to it a fair few times i was unimpressed. But today i knew it was what i needed and an hour later it gave me what i needed.

I must have countless plugs I've almost written off but that really i just need to learn properly.

[RobC]

Sometimes a transient, in my experience, sounded better with hard clipping, than soft clipping with rounded edges.

Or, maybe some sort or click-rejection? See the hard gaps? (I did that with Goldwave's amplitude shaper. It took anything above 75%, and turned it to 0 in this test. Then I normalized.) Normally when clipping a transient, clicks get pretty weak, and the sound sounds flatter. However, such gaps might add back what's lost with that edginess in form of click-length distortion-like effect.

Untitled.png (6.9 KiB) Viewed 1081 times