unusual male vocal compression

I am trying to comp a male vocal where the singer often switches between chest voice and falsetto.

I want to create as smooth a transition as possible, and am wondering if some sort of selective compression of the chest voice (which is almost always louder and “fatter” would be possible, and if so, how to do it!

I use iZotope Alloy/Nectar for my vocals, and also own Ozone.

Thanks in advance for any suggestions!


Duplicate the track and edit the chest and falsetto to playback on each separate tracks. Now each can be treated on it’s own.

Yes, split the tracks. I would also be levelling the individual comp events (event volume) as I go along, to smooth out volume changes before the inserts too. Then I would probably have two compressors (slow and fast maybe, season to taste) and automation on the track fader as well for final mix balancing. All of these tweaks might mean you don’t need to split the track in the end, but it’d really depend on how it all sounds in the first place. Quite often I track split verses from choruses for intimacy vs power say.


If this switching takes place in a structured way (certain sections or even sentences), the easiest solution would be to just record both voice-types separately. But singers are strange animals (I know cause I’m one) so this may or may not work…

Personally, I would keep it all on one track and automate levels. Separate processing makes it trickier and it’s easier to end up with something less organic. Usually there are scoops aand swoops between voice changes that would fall into a middle ground. Just level out the vocals by ear if there are substantial level changes, and see where you are then.

That’s why I said “IF this… etc”.

Common problem when navigating between head and chest voice is letting your voice crack too much rather than smoothing the transition and making the “crack” as inaudible as possible. Tough to edit that yodel out.

If it was me, I would use 2 different tracks if the qualities of the sound were drastically different. Automating a single track with 2 distinctly different sounds is a lot of unneeded work IMO because you can achieve a similar vocal treatment by duplicating the channel settings and tweaking the eq and compression for the two different tones.

If you are dealing with an experienced singer, a single track should work ok.

If you find yourself constantly tweaking a single track to keep it all together, switch to two.

If the falsetto is strategically sung, like in the hooks or turn-arounds, 2 separate tracks allow you to embellish that in a mix. You can also aux it too.

Good luck.

Thanks very much for the replies!

Luckily the singer is very adept at switching, so no “cracks” to worry about.

This is a pop/dance track, and I can afford to be quite “in your face”. At the moment I ask him to double the track at a lower pitch (say 4/5 semitones down) so that the entire melody can be sung in chest voice, then pitch it back up in Melodyne with a slight format shift and mix it in at half-volume which helps to support and lend “weight” to the falsetto parts. Now I’m wondering if there’s something beyond EQ I can do to just the chest voice “sound”.

I’m betraying my lack of knowledge here, but what would the two compressors be doing?

I was thinking I could maybe somehow use a multiband compressor the bite into the chest voice frequencies whilst leaving the falsetto intact, but I am not at all adroit at using one!

Ideally I would prefer not to split the track.

Hey lappinit, I don’t suppose there’s any way you could post a link to the before and after? Your description of double tracking it with a Melodyne+formant repitched lower-toned track sounds really interesting!

what would the two compressors be doing?

It’s a technique for smoothing things and also making them more exciting, someone recommended it to me a long while back, but it’s not a hard and fast rule. The basics are that you compress first very hard with say 12dB of reduction with fast attack to tame the peaks quickly, almost limiting in a way. This pretty much flattens everything out. Then to add some bounce back in, you compress 3-4dB after that with a slow attack and fast release. The slow attack then increases the dynamics by allowing some of the peaks back in before compression occurs, hence you get a smoother vocal overall but with some excitement included. That’s the theory I believe anyway.

Using a multiband compressor is also a useful modern day tool, but you would probably use that when there is a problem frequency which occurs occasionally during the performance. And maybe in this case it’d be better to automate an EQ change during the song, which makes it much easier to control.

I would say that it is now standard practice to automate much more to do the job (or even the artist!), simply because we have easy automation available. If you read some of the SoundOnSound articles about mixing famous songs you will see that they’re riding vocal levels throughout, to emphasise the words and the feel. It’s not just ‘set and forget’ mixing using just inserts anymore.

Finally, if you split the tracks (probably mentioned above) then you simply need to send both to the same group and you’re back with one track again. I do this sometimes for wildly different sounding takes - like AM/PM. So I’d split out, EQ one to sound like the other, then buss to a group where all the inserts are applied. It’s a pain but that’s the way it has to be to correct that sort of problem.

But there are no hard and fast rules :slight_smile: Hope this is useful.


What Gargoyle said.

Also, say you have a vocal that needs different EQ treatments for the verses and choruses. Insert 2 EQ’s in series and bypass the second one. On the chorus automate the bypass of the first one off and the second one on. Easier than automating 3-4 EQ parameters and you don’t need to split the track. You can do the same with comps by just using two instances and automating the bypasses. Easier than using two audio tracks.

This is why it’s good to have unlimited pre-fader FX inserts, so you can easily do things like that and still have available insert slots but when it comes up people say … “Who needs 8-10 inserts slots?” :slight_smile: Well, if you’re mixing and doing what I described in the first paragraph you might have some plugs doubled up in bypass on a single track like that.

This is also why naming plugin instances helps, so you can keep track of duplicated plugs, so you know what each is supposed to be for. So your second EQ will actually say “Chorus EQ” right in the slot.

Thread hijack… quick question. I thought that bypassing plugins caused problems because when you turn them back on again they replay their audio buffer from when they were turned off. Does this not happen anymore, have I just not noticed it’s been fixed?


Yes. You can get pops and automation data is not sample accurate.

Aha, OK, still there then. Always thought that clicks and pops are ugly, better to auto crossfade with a guaranteed no pop so we’re none the wiser, or bleep so we know something’s wrong, one way or the other :slight_smile:


Well that’s a bummer (a bug?) and it shouldn’t be happening.

Anyway, the bypass technique is valid in a properly working system without that particular issue but yeah, if the DAW is spitting out other stuff from a previous buffer when you toggle bypass, that would be a very good reason not to do that.

I usually (if possible, but not always) toggle my bypasses when nothing is actually playing through the plug. I can’t recall ever personally running into that particular issue.

Very helpful answers, thanks very much!

I am very guilty of this, need to get my hands dirtier with automation.

Will do! Tbh when I’m in a hurry I do the pitch shift in the box and don’t bother about the formant - the level is low enough that artifacts aren’t too apparent (depending on the mix).

It belongs in another thread… But… Due to using UAD plugins I often have a large plugin latency introduced and although I haven’t upgraded to the latest version yet I notice this problem a lot. I think the plugin processing is turned off when you bypass but when you put it back on it’s reactivated and I hear the old buffer emptying before it catches up with the latest input. When I get a moment I will upgrade to the latest Cb version and see if it still happens. If so, then I’ll raise a specific thread.

But most plugins have a bypass button, although you do have to be careful because if they’re hardware emulations then they may emulate the bypass too, UAD do that for authentic operation.

</Hijack - ended hopefully>


Here’s another thought, you might want to check out the Waves plugin called ‘Vocal Rider’. I demoed it here this week and it is much better than a fast hand.

Using automation for what you are doing, with hand editing after the fact, you might do well, hard to say. If the sections, full voice to falsetto are concise and of some length then you can do it with Steinberg automation, no problem.

Getting back to the VR, it has side chain capabilities. However here, unlike how the side chain works in a compressor, say you use a kick drum inserted into the sidechain to clamp the bass guitar, the VR behaves just the opposite. It instead raises the volume of the vocal vs. the instrument mix. I am pretty sold on it and will probably buy it and I am not afraid of automation. Just a thought.

Here’s another another thought! Had looked at that, but waiting for this to come out in VSTinstead: http://www.automaticmixing.com/ ,

based on this comparison with Waves Vocal Rider: http://www.soundonsound.com/sos/may11/articles/wave-rider.htm