I really love my Voice-over/ADR sound now. I use a TLM 103 amped by a SPL Channel One mk2 Premium (which I use for some 3-6 dB compression, gentle DeEssing, Highcut (50Hz) and Air-band boosting by 2-3 dB (14kHz) ). Usually I use a plug-in chain of Vocal Rider (2dB slow mode), CLA-2A (Opto-Compressor), SSL Channel (to compress the transients the Opto couldn't catch and removing some ugly mids around 800 Hz), Fabfilter-DS (DeEsser) and a very short convolution reverb if the mix needs a wet Dialogue.

I am extremely happy with this chain, but I have one big problem. Breathes. I read the other Q&A regarding the same question but it could not answer my problem since they told to use a noise gate, which does not work for my setup. Thats why I use manual volume envelopes (6-9dB before compression seems to work best) to reduce the breathes, but it is to time consuming for a lot of low budget productions.

I tried the Waves DeBreath at a friend of mine which just did not work as I wanted it too. Also I tried different kind of expanders and Noise gates, but they just don't seem to work for me, or I am to unskilled to program them. At the end I need to automate the parameters of the Gate/Expander which takes even longer than automating the Volume.

Do you have any recommendations? Or is there only the manual way?

  • If someone is interested. I measured my time. I need around 6x the playtime to clean up a dialogue. 3x to get rid of all the clicks, "ähms and ehhrrs" and unnessesary breathes (autofades help here). Then 1x to ride a fader to reduce the gain on needed breathes like the typical breathes after or before a phrase. And 2x for fine-tuning and final listening. Do you think that is ok time vise? Also today I had an Idea to fill the gaps of my cuts (to keep timecode position). On a Second track I use a room tone treated by the same FX shaun that I sdechain compress by the Dialogue -> It fills the gabs Mar 14, 2014 at 18:25
  • I generally don't fill with room tone if its a VO but if your using pro tools you can set 2 tracks to the same voice and when the master track doesn't have audio it will switch to the "slave" track. That way you don't have to place tone as precisely.
    – coaxmw
    Mar 15, 2014 at 1:08
  • Ah that is another interesting Feature! I will give the manual another go :D Mar 16, 2014 at 6:51

4 Answers 4


I tend to go manual. When I'm editing VO, its just a matter of a fast breath cutting pass and then a separate pass of other editorial and mixing.

For breath cutting, I'll make the waveform and the track very big so that I'm just looking at the softer stuff in the track (not worried about seeing peaks when breath editing) then i'll place my left hand on the right side of the keyboard with my middle finger on the delete key and my thumb on the left arrow key. I'll place my right hand on the mouse.

the workflow is then:

  • highlight breath with right hand
  • delete with left middle finger
  • center cursor with left thumb
  • repeat

by centering the cursor with your thumb after a cut you have the effect of advancing the timeline. If you need to advance further you can just click the mouse with your right hand and advance again with your left without any movement on either one.

I find this to be the most efficient way to cut breaths for me.

  • Do you use fades then? And how many breathes do you cut? I try to let a breath between phrases and remove breathes in-between a sentence. My manual tactic right now is highlighting the breath -> separate it -> lower gain of breath by -6dB -> crossfade in and out of breath. Mar 13, 2014 at 15:53
  • I use the autofade setting in protools and do manual fades as needed (which is pretty rare) If the edit is to have breaths in, then I use an expander to reduce them. works out fine. The trick in YOUR rig will probably to be to do the expansion before you run all of that compression and levelling. trying to pull them down after having gone through that chain would certainly be more difficult than if you reduce them before they hit it.
    – Rene
    Mar 13, 2014 at 17:05
  • I worked with protools for 4 years now and i did not see the autofade function! Thanks that will help to bring up my Speed! Mar 14, 2014 at 7:30
  • Wow i never knew that either! Thanks a lot Rene, great write up! Mar 14, 2014 at 9:48

How long of content are you working on? I usually just cut most breaths out during rough playback and fine tune any edits while clients are discussing other stuff. I would personally rather spend my time cleaning up the read at its source rather than tweaking gates/ expanders etc to guess whats happening. Every talent is different also, some have huge breaths, others very slight. If it sounds strange to cut them I might just copy some volume dips and paste them over the other breaths and fine tune during the next playback. It all goes pretty quickly.

  • It is often corporate movies of 3-5 minutes. And for most i have 2-3 hours for recording, editing mixing with music and ITU 1770 meassuring! That is why i cant go that manual often! But for stuff i have more time or claims i realy go deep in manual treatment! Mar 14, 2014 at 14:27
  • Another option would be to ride your fader through a pass so that you can write automation to bring down the breaths, doesnt take any longer than the playback pass and maybe a few minutes for touching up some spots.
    – coaxmw
    Mar 14, 2014 at 15:28
  • Yeah fader automation would be a possibility, but I am working digital right now, so I would need a good Midi-Fader. Can you recommend some? That could really be the fastest way. Mar 14, 2014 at 15:57
  • I'm not sure what program your using but I think you can get some pretty cheap, like a faderport or beringer. If i'm not on a control surface I will sometimes just write it with the floating fader window in ProTools, no other hardware required. If your using ProTools you probably would need to step up to an Artist Mix I think.
    – coaxmw
    Mar 14, 2014 at 16:43
  • Haha that faderport made me smile. Thats a funny gear :D I will check that in a lokal store :D Mar 14, 2014 at 17:57

so you have the SPL compressing, then the CLA-2A compressing, and the SSL channel compressing.. hmmm.. why not disable the SPL compressor, then gate gently (-6 to -10dB) with lookahead in your DAW, and only then have your compression take place. of course, volume automation is the best, but when pressed for time, a cleaner workflow might get a few of these problems out of your way.

  • I have to Test that! Even so i am realy in love with the spl compressor! The easiest well functioning compressor vor dynamic stuff like dialogue! Mar 14, 2014 at 7:34

I either manually edit them out using a method very similar to Rene's or if breathes are desired in the edit I will use clip gain to pull them down to taste. IMO there is no good substitute for taking the short amount of time build your dialog edit. On a 3min corporate video with VO, this should be doable in the time you mentioned to get the whole mix done. You will get faster the more dialog editing you do. There are tricks we use with region groups too when recording the dialog that makes the whole process quicker. Basically, in a few quick passes you should be able to get it locked. In short time you will know well what a breath waveform looks like and you can clean a voice track faster than you can play it.

  • Yeah! It seams that I have to do it the hard way, or if there is no budget, let the breathes in or just take out the obvious ones. Mar 14, 2014 at 17:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.