desser exmple???????

MegaHurtz · 9 Jun 2012

It`s de-esser.

aviadderi1 · 9 Jun 2012

But if I connect them together it's make a compress of all the signal and I need to compress specific freq

MegaHurtz · 9 Jun 2012

use a lowpass filter on your sound, and compress the corresponding highpass section.

or basicly a lowpass filter, and in-the lowpass filter to be compressed.

ChrisHooker · 9 Jun 2012

lol - ok, it kinda "hurtz" to read this, so I'll reply...

What you actually need is a band-splitter. (At least 2-band) To separate the lows and the highs, but not get rid of anything. Then you slap a compressor on the highs, and finally recombine everything.

As another option (probably what MegaHurtz was implying?), you could split the signal, with either a band-splitter or EQ, use the filtered signal as just a control signal to feed the envelope of the compressor, and then apply the compressor to the unaffected split. (Wideband compression based off of only the high freq's.)

RJHollins · 9 Jun 2012

If I may ... the descriptions given seem confusing ...

A 'quality' de-esser would allow a specified bandwidth to be 'extracted', and selectively compressed, then added back to the remaining unaltered signal. Visually it would be considered a 3 band split, with the 'middle band' able to slide.

As to design ... not all de-esser use compression as its functionality. There is another design to consider that uses a controlled 'phase inversion' level to be applied to the selected band.

Not telling anyone how to do this, nor supplying the definitive description ... just hoped to clarify. 😍

ChrisHooker · 10 Jun 2012

Of course you may, RJ!

...A nice elaboration on my elaboration (or at least what I intended to write, lol), plus a third method I hadn't thought of.

Although, I have to say - to implement the phase inversion technique, like you said, it'd have to be controlled. You wouldn't want to add a full-strength phase inversion to the original signal. That would completely cancel out. So you'd need to adjust the amount of it to add in. And it wouldn't be a constant amount, or that would just equate to an EQ. (No pun intended) If you want to adjust the dynamics, you still need an envelope control... and other parameters like onset speed and recovery speed (attack and release) that really just bring you back to a compressor. So I don't readily see the benefit (or much design difference) of applying phase cancellation as opposed to volume control for this purpose.

Could you go a little more in-depth to explain? I find this stuff kind of interesting.

RJHollins · 10 Jun 2012

Hello ChrisH,

I was just expanding on your post ...

but as to the 'phase invert' de-esser ... hmmm, I can't provide much in the programming aspect, but I can describe what it is I think it is 🙂

First ... the 'handling' of the de-esser process DOES seem different comparing band limited compression, and the phase invert variety.

There is only one plugin [from Steinberg - I believe - need to verify] that uses the phase invert technique. [that I'm aware of].

From how it seems to work ... YES ... it is definitely using a 'level' control [or maybe better put, an amplitude control] for the inverted phase band that is added back to the original signal. What I do NOT see [in the controls] are any of the normal compressor settings, such as, threshold, ratio, etc. It 'almost' seems to be acting as a type of 'dynamic EQ', that with its' inverted phase is 'control nulling' [my term] the original signal ... does that make sense 🙂

From what I recall, the band is fully slideable, and so is its width [almost like a 'Q' setting from an equalizer].

Whenever I use this design of de-esser, it tends to work nicely in my mastering work, where basically we are dealing with 2-trk masters. This design can be rather transparent, as the issue of 'pumping' or 'ducking' doesn't really occur as with band restricted compressor designs. Granted, its nice to have a choice in strategies 🙂

One point of interest [for me] is that the de-esser tools that are out there [and I do own several plugins of them], seem to be very under developed in their GUI display. If you look at products from guys like FabFilter, and their GUI for their compressor & limiters, they seem to exploit the availability of having a computer monitor to SHOW you what is going on.

It seems the de-essing tools should try to help identify the [more often than not] several problem areas of a vocal track [for instance]. We could have all types of sssss, whistles, vocal harmonics, etc, that often would like to be tamed.

Adding to the difficulty can be the rather short burst of the problem. I mention all this in hopes someone will develop THE de-esser plug 🙂

ChrisHooker · 10 Jun 2012

Here's som'n I pieced together today. It's a multi-use single-band compressor (wide-band or split).

I haven't added any preset handling or automation, nor checked/listened to see if there's a processing delay/latency (which could be heard with the mix not set to full wet). (It's late here, so I can't listen right now.)

It's a work in progress, but it does include an FFT, as it sounds like you were alluding to, RJ. Also, it has the capability to process a select range of freq's, not just all highs above freq x.

It might be a tad confusing/complex for trying to decipher (not a very clean schematic), but maybe it'll help you see how it CAN be done, Aviad.

trogluddite · 10 Jun 2012

RJHollins wrote: This design can be rather transparent,

The phase invert can also have extra transparency because when the 'ess' signal is below threshold, the input signal is completely unchanged. In a normal 'split band' design, the main signal is always passing through the band-splitting filters and then being re-combined. Depending on the choice of filters, this might lead to phase shifts that subtly alter the tone of the sound, even when the de-esser is 'doing nothing'. In multi-band mastering compressors, this is less of a problem, as they will not be used 'live', so minimum phase shift filters (which introduce latency) can be used to minimise filtering artefacts. With the 'phase invert' method, only the side-chain needs filtering - to achieve gain reduction, the side chain is inverted, gain-adjusted, and added back to the main signal. So when no de-essing is happening, there are no filters etc. in the signal chain at all, just the pure input signal. It could also possibly make the design simpler, and lower CPU, as the filtering only needs to extract the central band-pass area of the sound - no need to 'extract' the high and low bands as you would in a true multi-band compressor.

ChrisHooker wrote: Here's som'n I pieced together today.

Thanks, I will have to play with that, it looks very comprehensive - and the GUI concept is great too. Little touches like the threshold that overlays the metering make dynamics so much easier to understand.

ChrisHooker · 10 Jun 2012

Sounds cool, thanks for the insight, Trog.

Also for the intro kudos on the compressor... I just noticed, though - the mix knob needs to be reversed to a range of 0-1, instead of it's current 1-0. I've just updated the osm posted above.

MegaHurtz · 10 Jun 2012

Ofcourse people are going to try to make it. But ive had better result getting rid of cheap mike essing, using just a lowpass filter (decent eq) and (mandatory compressor). instead of compressor on highpass section. But that may be a methods thing.

ivanschrago · 15 Aug 2012

Here is an example I made for myself some time ago to compare with waves de-esser. I used many parts from the forum, so please don't kill me for not givin' credits to all of them, cause I really don't who created most of the modules. 😇

Not perfect de-esser, but has switchable bandpass and hi pass filter. so I think, that is how a de-esser works.

greets