Hey guys I'm working on a short film and I am wondering what kind of techniques people use to help get rid of noise floor or room tone. I usually use X-Noise and EQ but I'm wondering what other plugins people may use or techniques. I ran into the problem of low dialogue on one recording and high on the next so when I bring up the dialogue on the low it brings up the room tone or noise floor and doesn't match the higher one at all. Let me know what you guys use or how you go about this?? Thanks!!!

  • also depends on the shot, if the low dialogue is on a wide, steal audio from the MS or CU and add some verb to taste. Commented Mar 20, 2012 at 23:46

8 Answers 8


Very kind of you Brendan to link to the blog post, glad people are finding it useful!

As Melissa has touched upon (and that book is a very good introduction to dialogue that I equally recommend), and as briefly mentioned in the blog post, editorial is the foundation here. No amount of Broadband noise suppression, even a Cedar DNS, can fix a dialogue track which isn't properly prepared from the ground up.

This entails doing any and all edits necessary to make the dialogue track play smoothly and seamlessly (as humanly possible) in raw form without a sidechain. It entails carefully building out backfill to both get rid of unwanted noises and create very long, gradual fades where necessary to 'hide' the noise bumps of different angles. Sometimes the required editorial to whip a dialogue track into shape can be downright intense and feel nearly impossible (and even frustrating when you have only 5 seconds of clean backfill to work with sometimes).

For dialogue editorial, even before running a sidechain, I personally like to take a 3-Pass Approach:

Pass 1: Down-and-dirty cleanup, brute force edits to make sense of the OMF, remove major offending noises, split out all PFX, backfill where necessary and smooth all edits, level out all my edits' gain for a consistent -18dbFS playback level across the show (= to about -27 dB LEQ(A) for Dolby LM100 or -24dB LEQ(a) on a Stereo Phase Scope I believe) - work at a pace of 2 minutes of footage/hour to create a rudimentary dialogue track which is 'temp-mix ready' if it hand to be shipped at a moments notice. Think of this as the organizational and "cover my ass" stage ;)

Pass 2: Thorough Deep Cleaning, removing every single lip smack, tick, pop, click, crackle, plosive there is to be removed - I prefer the manual labor approach rather than signal processing. Anything which is processed at this stage has the original source edits copied down to X Tracks

Pass 3: This is where I swap in all the required alt takes, phonetic and phrase fixes for items I noted along the way like Overlaps, Swallowed letters (such as not pronouncing a consonant like 'T' at the end of a word), and other Noise-related issues (maybe a low-level or intelligibility issue) - basically, anywhere in which dialogue was compromised, this is the Phase where I make sure we have all of those fixed by way of the dailies. All original edits being replaced are copied down to the X Tracks.

Once this is all dialed in, then setting up a sidechain with something like a C4 can help you suppress the background noise, which through all of this preparation you have made consistent in level and timbre so the C4 is working FOR you not AGAINST you. And, at this point you'll be able to properly evaluate material as you roll over it, and maybe it needs a 'touch' of FFT like iZotope RX or X-Noise to make it sparkle. Now would be the time to do it, while making sure to keep a copy of the original on your X Tracks. These X tracks are 'cover your ass' tracks for the stage, especially if you're doing your own premix pass before it hits the stage (as they will likely want to tweak it, sometimes undoing or altering some of the rendered processing you did, so the X Tracks are a must).

Also food for thought: physcoacoustics are a powerful ally. Sometimes we add noise to remove noise ;)

EDIT: Here's a link to another post about what's involved with the above-listed 2nd Pass: http://www.stavrosound.com/blog/wordpress/2012/02/the-rice-crispies-mix-snap-crackle-pop/

  • @Stavrosound Thanks for the excellent breakdown. When leveling gain, what technique do you use? Track volume automation/audiosuite etc? Commented Mar 22, 2012 at 21:41
  • Sure thing! Actually, I only used AS udner the most rarefied of circumstances. I level-out in track gain automation. However, if the dialogue does not naturally sit around -18dB when an unity gain across the show, I'll load up non-automated RTAS trims on each track with an identical gain value to serve as a "corrective re-calibration" so that my automation faders remain at unity with a 1-8dB level, that way I can make the most of the automation resolution and have enough headroom. Commented Mar 23, 2012 at 0:42
  • In 15+ years of editing that has to be one of the best descriptions of the process I've seen! Well done!
    – Sonsey
    Commented Mar 23, 2012 at 15:40
  • Epic Stavro. Nice one.
    – lucafusi
    Commented Mar 23, 2012 at 17:14
  • @Sonsey Thank you for the generous compliment. Thanks @Luca too. Again, only one way of doing it, but I've found it to work effectively for my own workflow. Hopefully others are finding it helpful/useful. Commented Mar 23, 2012 at 17:17

Stavrosound linked to a great post he did on the Waves C4 being used for cleaning dialogue. Here


John Purcell, on his book Dialoge Editing, advises strongly to have a proper edit regarding the room tone before you run to the noise reduction. Having longer fades on room tones to trick our ears between aggressive transitions; If using plug ins like x_noise or alike, and turn to destructive, maybe try some eq first, know the noise offending your audio so you can work more surgically on it. Good luck!


Is the shot with low-level dialogue the only unusually low recording? Are there alternate takes at a more acceptable level? Using either one of those, or audio from another shot entirely like Michael Gilbert said, you can likely edit together a line that sounds convincingly enough like the original's tone and performance. Make sure that you've edited your dialogue fill so that you have sufficient handles going in and out of each shot, this makes it easier to sneak it in and out.

If you can't cut an alt and don't have any luck with noise reduction/C4, best thing I can think of would be to take that loud fill from the quiet line, cut together a looong piece of it, and extend it over most of the scene.


iZotope RX


Used it once wish I had the money to buy the advanced version!!

  • The non advanced version is fine for that if you have the money. You could try some multiband expansion to get done decent results too.
    – Andy Lewis
    Commented Mar 20, 2012 at 16:08

iZotope RX2 is excellent for the money.


@stavrosound awesone answer, please check out my question about post production. I ususlly always do your phase 2 deep clean process to the commentaries that i work on, which can take me up to 3 hours or so.

To the asker of the question, one good thing to do might be to much the silent parts of the dialogue via automation.. only level up the volume on the audible part. Too much fx processing to the whispering or quiet tones will mske them sound unatural. If possible it might be a good option to rerecord the actors or actresses.

  • Phase 2 takes a long time, what you experienced is normal. Usually Pass 1 takes 1-2 days per reel, Pass 2 takes 3-5 additional days per reel (depending upon the problems, pass 3 can take anywhere from 0.5 days-3 days (depending upon what needs to be addressed, and subsequently how much material has to be prepared for X tracks and alts properly) Commented Jun 21, 2012 at 8:50

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.