DeEsser before or after normal compressor?

Hi there guys.

As the question states: would you use a DeEsser before or after the normal compressor on a vocal chain?

I am mixing a female voice here and I just can’t seem to decide.
She sings very good but I need to compress her quite hard to make her fit in the mix, and as I do so the sibilants come sneaking in.

Any thoughts on how to go about this?


I have not had much success using deesser. My internal logic says to deess before compression assuming the deesser works better on the uncompressed signal. I have been able to identify esses and Ps on the waveform in the editor at the start of words and have used selective gain reduction successfully. 2, 4, 6, up to 10 db in rare circumstances. You could try that on a duplicated track.

A little bit before, and a little bit after the compressor.

PS. Remember to de-ess the reverb send and/or return (often a source of “where does this ssss’es come from”).

That’s what I do, but it’s not rare for the in-the-middle-of-the-word “ssss” to cause adv problem that way. Any hints?

I’m going to try this. For middle of word, if I find one, I’ll try selective deessing in the editor.

I use a de-esser at the end of the chain…

You’re potentially adding in more sibilence with inserts added after the de-esser so I put it end of the chain. Gets the job done for me, happy with the sound i get.

It’s one of those things though… I dont think it matters a huge amount as long as it works.


i figured this out a little while ago… it’s can be much kinder and more natural sounding, particularly on heavier sibilants than using a single de-esser.

This is why the channels need more insert slots. If I’m working on a lead vocal I create a buss for it immedietly because as soon as you’re doing things like having two de-essers you’re running out of inserts. A work around would be to use the racks but I’m not a fan.

Thank you for all your tips friends/people.

But I will never go in and edit an audio file to get rid of sibilances.

For me it is ok to treat a take with efx but not to go in and alter it at sample level. If you do: then it is not that same take anymore


As an afterthought… if you have melodyne it’s really easy to just turn them down a little too… sounds completely natural if you use your ears.

Of course 8 or 12 inserts would have been better.
But from that to say Cubase sucks and I can not do my biz because Cubase only have 6 real insert points per channel?
Nah… That is just stupid, and have you heard about say… routing from group to group… Means you get 12 inserts. Is that enough for ya?


I didn’t say it sucked? I love using Cubase - I’ve used it for 16 years… and in the post you quoted from I mention routing the track to a buss…? :unamused:

Keeping an unaltered digitized wav could prove benificial but once exported or bounced with effects, regardless at what stage applied, the resulting wav is altered then altered again to reformat. Surgery in the editor has been fun and effective for me at times. Like Dr. Frankysteen I’ve created a few monsters but I can just make another copy of the original digitized wav and drag it back to the laboratory. :nerd: