How to apply an effect only during specific spans of time in a track?

Question

I am working with some voice over that will be played quietly in the background, and am finding that even subtle sibilance, the sharp hiss or "ess" sound made when "S"s and "C"s are pronounced, is distracting when played like this.

The DeEsser tool in Adobe Audition is certainly easy to apply and does provide a perceptible improvement at normal listening volume. However, at low volume for background listening it makes no noticeable improvement to the sibilance, and comes with the cost of adversely affecting the sound of the whole VO.

I have found that my best results have come by simply de-amplifying the track at the sibilance spikes. This is very time consuming, and I want to automate it or apply an affect that does this.

Is it possible in Audition, or any piece of software for that matter, to automate the process of finding sibilance spikes then applying an effect only during such spikes? To turn an effect (i.e. multi-band compression) on or off - up or down, as the track plays? Can I find and flag a frequency profile (a sound), then listen for that sound and when it is present, turn on an effect?

Thank you, in advance, for your assistance.

Answer 1

If you're looking for a quick way to process the whole VO track to attenuate these sibilance sounds, the most simple way is to check where these sounds are most prominent in the upper frequency spectrum by using a sweeping positive EQ peak then attenuate it to a satisfactory level. It shouldn't affect the rest of the audio much, as these sounds tend to be most noticeable in the upper vocal range and even beyond.

You could take it further and (as Mark commented) target the frequency/range with multiband compression so you can still keep the unaffected audio in the targeted range. That's basically all a classic de-esser does, but this way you have more control.

Answer 2

I will provide here a kind of automation for the editing of sibillance in a vocal track.

I will not cover any specific program cause every program is a bit different but generally I'll be talking about industry standard programs like cubase/studio one/pro tools etc..

1. Add a gate to the vocal track. That gate should band pass all the other frequencies and target only frequencies above ~2KHz and below ~10KHz (YMMV). Set a quick attack so it opens quickly and set hold to 0 and release to something feasible but not too long (20-30ms? YMMV). Now this channel should only open when there is signal in that band and above the set threshold. If your signal is too low on volume and the threshold is never surpassed then add a mixtool/utility/gain tool right before the gate until you have a healthy signal which does what we want.

2. Add another channel (should be stereo in most programs i know) and feed the gated vocal channel (step 1) as input and hit record. If you've set the settings correctly the new track should print the exact positions of the sibilance sounds you've gated from the original signal.

3. Now you have many ways to go on about using this printed channel.

   1. First and most obvious use it as a guide/map and edit out the sibilance yourself cause now we're confident that these are the areas that cause the initial problem.
   2. Flip the phase, in many occasions just flipping the phase and playing with the fader causes the right cancellation and you end up with something that works
   3. Feed the produced track as key to a multiband-sidechain compressor that only works in the high or even better band passed area of the signal. Also be sure to stop routing the produced track to the main mix so you don't listen to the sibilance from 2 sources or even better set the sidechain send to pre-fader and pull the fader down to 0. This track can now be as high in volume as you want since it only feeds the multiband compressor.

There's a lot of fiddling in this process to get it to work right. I would choose the simplest way of editing the sounds my self using the printed (map) channel, other ways are a bit more elaborate and also depend on how proficient you are in audio mixing and manipulation.

There are some shortucts especially to gating the audio, like using a de-esser plugin which offers audition of the signal that it works on, which essentially is the same as the gate.

Generally if you have a signal that is too low on amplitude and it wont hit the threshold you can always, in nearly every dynamics processor, crank up the input volume - let the compressor/processor work - then turn down the output volume to taste. Even if the plugin doesn't offer that you can easily do it by using 2 stock utility/tool plugins that all programs offer and add one before and one after the processor to control the input gain and the output result gain.

I hope that helps you or someone else, good luck! :)