Your Favorite De-essers?

[Ruzz]

Originally Posted by John Spence

Volume fading will not reduce the sibilance, only turn down the vocal for a millisecond or so and may lead to some very unnatural artefacts in places where the voice is exposed but if it works for you...!
Happy new year.

Hi John - is there some science behind this quote? If so, what is it?

As I understand it, a de-esser affects the frequency spectrum of a track in a non-linear fashion (a fine-tuned compressor). A fader, on the other hand, automated to attack troublesome 's' sounds will affect only those parts you choose to attack.

When I automate a fader, I can hear and see the problem wave. It's precise. I can see where the bulk of the energy is in the waveform. No doubt a plugin can do the same, but to achieve its results, its calculations are going to colour the waveform - surely? And if the plugin is not automated ("set and forget") it's going to do that to the entire track affecting to some degree the air and presence parts of the signal too (depending on settings, of course).

An automated fader doesn't do that.

Finally, there is no voice present during 's' sounds (the most troublesome sibilant). An 's' is easy to isolate visually in a waveform. When you try to home in on a set of frequencies with a de-esser, you're relying on getting a bunch of variables correct and (worse still) those variables are constantly changing - it's a moving target - no one 's' is the same as any other. Whatever settings I settle for, I'm always left wondering if those settings are the best they could be.

Using an automated fader, I'm looking at a fixed piece of waveform with just an 's' sound. Easy.

I'd just like to know why you think turning down the volume of a sibilant doesn't turn down the volume of that sibilant... and what those unnatural artefacts are that you mentioned.

Ruzz

[John Spence]

[QUOTE=Ruzz;82389]Hi John - is there some science behind this quote? If so, what is it?

Hi Ruzz,
You raise some interesting points so here goes!
By unnatural artefacts (artifacts?) I meant anything which makes the voice content sound unnatural.
Let me say now I have never tried to De-ess a vocal track by volume fading the sibilant parts so my observations are based on my own experience. After all this thread is called "What's your favourite de-esser".

[QUOTE=Ruzz;82389]"As I understand it, a de-esser affects the frequency spectrum of a track in a non-linear fashion (a fine-tuned compressor). A fader, on the other hand, automated to attack troublesome 's' sounds will affect only those parts you choose to attack."

A De-esser should reduce only the (user selected) frequency band.If you strap an outboard equaliser across the side chain and you can listen to the De-esser key then it's possible to very accurately home in on the offending area. By adjusting the threshold and ratio by which the de-esser compresses these frequencies then excess sibililance can be reduced to a more acceptable level. Obviously careful listening to the whole program is required to make sure that other parts are not compromised sonically.Occasionally I end up splitting a vocal track down multiple channels and having different de-ess and compression settings for different parts of a song but I'd much rather spend time at the recording stage and head off these problems.
The de-esser will reduce certain frequencies and leave what's behind intact.

Originally Posted by Ruzz

"Finally, there is no voice present during 's' sounds (the most troublesome sibilant). An 's' is easy to isolate visually in a waveform. When you try to home in on a set of frequencies with a de-esser, you're relying on getting a bunch of variables correct and (worse still) those variables are constantly changing - it's a moving target - no one 's' is the same as any other. Whatever settings I settle for, I'm always left wondering if those settings are the best they could be."Ruzz

I disagree. There is loads of voice present during 's' sounds.The hard consonants are the very sign posts, the flags, if you like, which make sense of our speech. If you use a de-esser you can reduce the intensity of the sibilant frequencies and leave what's underneath loud enough to make the voice intelligible. A programmed fader will simply turn down the entire vocal content (albeit for a short time) which, possibly occurring many times during a song, may become apparent.
To be honest, if I was mixing something I'd recorded and I had to work through the whole vocal track and selectively reduce the volume of sibilant components then I'd think I had made a pretty poor job of the original recording.However, as I said in an earlier post-if it works for you....

[Audio~Geek]

Try it out John. You may be surprised.
Zoomed in to the right amount you can see sibilance extending past the rest of the word.
With volume automation you can take it down 6 dB or as much as needed and make it sound more natural.
It's just one technique of many for vocal/dialog editing.

It can take more time but is really a more natural approach because you're only changing gain, rather than inserting a processor.

[tiberius]

I use the one in the Focusrite Voicemaster Pro and it works very well.

[John Spence]

Originally Posted by Audio~Geek

Try it out John. You may be surprised.
Zoomed in to the right amount you can see sibilance extending past the rest of the word.
With volume automation you can take it down 6 dB or as much as needed and make it sound more natural.
It's just one technique of many for vocal/dialog editing.

It can take more time but is really a more natural approach because you're only changing gain, rather than inserting a processor.

Fair enough, I'll give it a go the next time I have both the need and the opportunity to do so and will give it my best shot!

[Ruzz]

Originally Posted by John Spence

A De-esser should reduce only the (user selected) frequency band.

See John? That's the problem - "should" and "user selected". I'm saying for this particular job, the de-esser solution is introducing a level of complexity that's unnecessary.

If you strap an outboard equaliser across the side chain and you can listen to the De-esser key then it's possible to very accurately home in on the offending area.

True. No argument there.

By adjusting the threshold and ratio by which the de-esser compresses these frequencies then excess sibililance can be reduced to a more acceptable level.

So it's possible but not inevitable (that the problem is reduced). I agree with that too.

Obviously careful listening to the whole program is required to make sure that other parts are not compromised sonically.

I step I don't need to take using a volume curve. You're also highlighting one of the points I was trying to make - that this can happen with a de-esser and the user needs to take care.

Occasionally I end up splitting a vocal track down multiple channels and having different de-ess and compression settings for different parts of a song but I'd much rather spend time at the recording stage and head off these problems.
The de-esser will reduce certain frequencies and leave what's behind intact.

Needing to use multiple de-essers highlights the problem most graphically. What you say amounts to saying it's a moving target that is not easy or straightforward to pin down when approached from a frequency perspective.

And John, you're talking about one single vocal track - this approach is simply making matters worse. A targeted reduction of gain is much more accurate and (I'd say) much less problematic.

I disagree. There is loads of voice present during 's' sounds.

Let's be clear, by 'voice' I mean a release of air travelling through the larynx causing the vocal cords to vibrate and a sound to be heard.

An 's' sound contains no voice.

Said another way, if an 's' sound contains voice then it's not an 's' sound - perhaps 'z'? I have very little trouble with 'f', 'th' as in 'think' (when voiced these are 'v' and 'th' as in 'the') hence and why I referred to the the most troublesome 's' sound - as in 'sassy'. If you want to get a serious headache, check out a good article about sibilance (in the non audio realm) - wikipedia is a good place to start.

A programmed fader will simply turn down the entire vocal content (albeit for a short time) which, possibly occurring many times during a song, may become apparent.

Trust me, this is not a problem I have ever encountered. You'd have to create a pretty shitty automation curve to cause anything like the problems you think will/may occur. I really think you should try this, John. Zoom in and have a play.

And btw, "simply" turning down the vocal for the duration of an 's' sound is exactly what I want to achieve. Using a fader, this is precisely and easily achieved. I really don't care (nor do I need to take account of) what frequencies are being targeted - the tool I'm using doesn't need (or care) to know. Whereas, using a de-esser, I'm introducing a set of problems I need to solve because I'm using a de-esser!

To be honest, if I was mixing something I'd recorded and I had to work through the whole vocal track and selectively reduce the volume of sibilant components then I'd think I had made a pretty poor job of the original recording.

It's not always the recording or a bad recording that's the cause. You're dealing with the entire sonic profile of all of your equipment in your space using your ears on that day - how many variables? Some of those constituent parts can accentuate (I mean unnaturally so) some of the frequencies you (we all) hear. In this case we're considering how best to address that one problem. I am suggesting that a focused attack on the exact part of the waveform that is at the problem source is better than any broad brush approach - which, even when set "correctly" - whatever correctly is - is exactly what a de-esser is. It doesn't matter how long you spend twiddling knobs, it's broad brush and "your best guess" as to what is the best setting (after all, to prove otherwise, you'd need to wade through every possible/viable setting, just in case ).

Sorry, I don't see how a de-esser makes this job easier or better.

But yes, an interesting exchange of views - exactly as it should be on recordingreview

Ruzz

[John Spence]

An excellent reply Ruzz, you are a very patient person!
I'll certainly try your method and will set up a de-essed version alongside to A/B when the opportunity arises.
In the meantime let me ask you this. Why do you think that a de-esser is adding some level of complexity? I don't understand what you mean by that.
Second question. If you were recording before the advent of the very sophisticated automation software we now have, how did you used to de-ess your vocal tracks and were you happy with the results?

[acoustic]

first of all russ john I have had this prob with a few people not only the ssss but the ppps or pops a pop filter took care of the pops but IF the lady who was speaking new how to hold the mic no pop filter would be needed and so I tried to adjust the mic to reduce the s sound and I found if you sing flat across a hand held mic the s is reduced but if the sss are caused by bad cords or electrical equipment this will not work------acoustic

[Ruzz]

Originally Posted by John Spence

An excellent reply Ruzz, you are a very patient person!

I'm sure my wife wouldn't agree

In the meantime let me ask you this. Why do you think that a de-esser is adding some level of complexity? I don't understand what you mean by that.

Maybe complexity was too strong a word, but, since I used it...

Couple of things:

1 - That de-essers, in taking a frequency-based approach, force you to take a bottom-up (vague?) approach to target the problem requiring a set of (very subjective?) steps to home in on the correct settings (frequency, bandwidth, ratios, etc)

Automated curves are a much more top-down approach. I just want to reduce a certain sound I hear and - more to the point - one that I can see and pinpoint directly on the waveform with ease. I couldn't care less what frequency components are present. I just drag that portion of the wave down.

2 - I was referring to the steps you needed to take for the more problematic examples you mentioned (multiple de-essers in the chain etc.) and then the careful listening to make sure nothing else had been unduly affected (which it very possibly has though it may not be audible in which case it's moot).

Second question. If you were recording before the advent of the very sophisticated automation software we now have, how did you used to de-ess your vocal tracks and were you happy with the results?

I've been happy with what the Waves plugin does, John. It's the method itself which I think is not necessarily the best approach.

John, I'm sure you're familiar with the phrase, there's more than one way to skin a cat. I see the de-esser as trying to keep the cat alive , claws intact while you try to skin it with a pencil eraser.

Whereas the automated fader reads: First, kill the cat. Second, skin the sucker. Who cares how?

OK, not a perfect analogy - but I thought it was funny

Rz