Hello!

I usually can understand signal routing descriptions with repeated reading, but I'm afraid I need to ask for some help on this one ... can someone please clarify the following passage from Bob Clearmountain referenced in Mike Senior's book (page 206)? I understand putting a fast attack/fast release compressor on the multed channel, but what is the "loop", and how is it "linked" to the original channel? And when he says he "turns on the dynamics" on the original channel, I'm afraid I get totally lost.

I suspect the main problem I'm having is that I can't picture what he means by "linking" the mult to the original channel ...

Thank you!


**************************
"Bob Clearmountain explains a particularly ingenious manual side-chain implementation that also finds favor with Jon Gass and Manny Marroquin: 'I mult the vocal to the channel next to it, on which I'll roll out all the midrange and boost around 7 kHz as far as it will go. I'll then send the output of the fader (my question here: is he referring to the fader on the mult?) into it's own buss and select a compressor on that channel to pick up on that buss, so basically it becomes a sort of loop in the channel (my question here: what is the "loop"? I'm reading this as a send on the multed channel to a comp, but I'm afraid I'm not reading it correctly because that's not what I would tend to describe as a "loop") . I put the compressor on fast attack and fast release, and link it to the original vocal channel (my question here: how is he making the "link" between the original channel and the mult?), on which I turn on the dynamics (my last question here: "...on which I turn on the dynamics" - is he referring to the comp described in the previous bit, or an entirely separate, second, comp?) . It works like a side-chain on the compressor and has the effect of a fast high-frequency limiter, its threshold being controlled by the fader on the parallel channel. It's really effective, very fast, and it de-esses very smoothly and much better than any other de-esser I have ever used. Plus, you can automate the threshold.'"

His wording is a bit odd, but I think he's sending the mult to a buss, and sending that buss to a compressor's side-chain input. The compressor is actually acting on the vocal channel. I guess the reason for him sending the EQ'd mult to a bus first is so that fader movements on the mult change the level of the mult relative to the compressor's threshold (if he were using an insert send or a pre-fader output from the mult, then moving the mult's fader would do precisely bugger all).

Cheers!

Chris

Seems pretty straight-forward to me and describes a fairly routine process that we all used to use, when no dedicated de-esser was available, or the de-esser available was not up to snuff. The same audio is fed to two channels, one is left open and not EQ'ed and the other is muted and 7kHz (or whatever the sibilant frequency might be) is boosted and the others are cut. The desk compressor (or an outboard) is activated and this is is used as a side-chain to govern the compression on the non-EQ'ed channel. Now every time that the vox sounds the SSS sound, this is boosted on the muted channel, the compressor kicks in and the 'live' channel is compressed.

In today's ITB world, this kind of trick is hardly ever used any more.

I think he knows what he meant... The wording doesn't make it easy, but the principle is fairly familiar... and I presume he's talking about using a large console with dynamics available on every channel.

Quote:

---

I mult the vocal to the channel next to it, on which I'll roll out all the midrange and boost around 7 kHz as far as it will go.

---

The vocal is duplicated to two adjacent channels. The first feeds the main stereo mix bus in the usual way, while the other is purely used to generate the de-essing signal -- and that one has the MF cut and HF boosted to emphasise only the sibilance element.

Quote:

---

I'll then send the output of the fader into it's own buss and select a compressor on that channel to pick up on that buss, so basically it becomes a sort of loop in the channel

---

The processing channel is routed to a spare bus and the compressor in that channel set to pick up its side chain input from that bus too. This just means that the compressor's side chain level (ie, threshold) is controlled by the processing channel's fader. The 'loop' is the processing channel output into the processing channel's compressor side chain input.

Quote:

---

I put the compressor on fast attack and fast release, and link it to the original vocal channel on which I turn on the dynamics

---

The processing channel's compressor is set to react as rapidly as possible -- and it will react most strongly to the heavily emphasised sibilant frquencies. The compressor is set up as a linked (stereo pair) with the vocal channel's own compressor -- which is also switched on (of course). This means that whatever the processing channel's compressor does, the vocal channel's compressor does the same thing -- because it has linked sidechain control signals. So, effectively the vocal compressor is being remotely controlled by the processing channel's compressor.

Quote:

---

It works like a side-chain on the compressor and has the effect of a fast high-frequency limiter, its threshold being controlled by the fader on the parallel channel.

---

So basically, all he's done is use the second (processing) channel to create a sidechain equaliser to emphasise the sibilant frequencies, and configured the signal path around that channel so that he can also use that channel's fader as the threshold control. The same thing could have been achieved with a conventional compressor in the vocal path, with a big dollop of side chain equalisation... but this way looks a lot cleverer ... and is probably more convenient and flexible.

It's a fairly common trick with large in-line consoles (big SSLs and the like), actually, and can be used to create frequency-conscious dynamic control or all varieties, or where the threshold of the dynamics needs to be automated -- dating back to the days before ultra-flexible DAWs.

hugh


--------------------
Technical Editor, Sound On Sound

Quote Hugh Robjohns:

---

I think he knows what he meant... The wording doesn't make it easy, but the principle is fairly familiar... and I presume he's talking about using a large console with dynamics available on every channel.

Quote:

---

I mult the vocal to the channel next to it, on which I'll roll out all the midrange and boost around 7 kHz as far as it will go.

---

The vocal is duplicated to two adjacent channels. The first feeds the main stereo mix bus in the usual way, while the other is purely used to generate the de-essing signal -- and that one has the MF cut and HF boosted to emphasise only the sibilance element.

Quote:

---

I'll then send the output of the fader into it's own buss and select a compressor on that channel to pick up on that buss, so basically it becomes a sort of loop in the channel

---

The processing channel is routed to a spare bus and the compressor in that channel set to pick up its side chain input from that bus too. This just means that the compressor's side chain level (ie, threshold) is controlled by the processing channel's fader. The 'loop' is the processing channel output into the processing channel's compressor side chain input.

Quote:

---

I put the compressor on fast attack and fast release, and link it to the original vocal channel on which I turn on the dynamics

---

The processing channel's compressor is set to react as rapidly as possible -- and it will react most strongly to the heavily emphasised sibilant frquencies. The compressor is set up as a linked (stereo pair) with the vocal channel's own compressor -- which is also switched on (of course). This means that whatever the processing channel's compressor does, the vocal channel's compressor does the same thing -- because it has linked sidechain control signals. So, effectively the vocal compressor is being remotely controlled by the processing channel's compressor.

Quote:

---

It works like a side-chain on the compressor and has the effect of a fast high-frequency limiter, its threshold being controlled by the fader on the parallel channel.

---

So basically, all he's done is use the second (processing) channel to create a sidechain equaliser to emphasise the sibilant frequencies, and configured the signal path around that channel so that he can also use that channel's fader as the threshold control. The same thing could have been achieved with a conventional compressor in the vocal path, with a big dollop of side chain equalisation... but this way looks a lot cleverer ... and is probably more convenient and flexible.

It's a fairly common trick with large in-line consoles (big SSLs and the like), actually, and can be used to create frequency-conscious dynamic control or all varieties, or where the threshold of the dynamics needs to be automated -- dating back to the days before ultra-flexible DAWs.

hugh

---

Thank you everyone!

re Hugh's wonderful breakdown:

1) In the processing channels send cascade, would the 1st slot be the group that routes back to the sidechain, and the 2nd slot be the comp itself?

2) I've never "linked" two channels with a comp before, though I've seen that switch on my Fairchild - I'm thinking each channel would send to the Fairchild, where they are then linked ... but if that is wrong, I'd be grateful for any correction please!

3) Finally - Hugh, can you expand on the added flexibility this might bring compared to highly eq'd side chain signal in the comp on the main signal path you mention ... and (bottom line?) is the result likely to be better than the stock de-essers I have access to in Cubase?

Thank you again, Hugh, The Red Bladder, and Korff!

--------------------
Alexis -Cubase 6.5.0/SX3.1.1.944, XP SP2, 4GB RAM (1GB not accessible, but used just to balance the computer so it doesn't tip over); Delta 66 in Omni i/O Studio; Motif8; UAD-1

Quote alexis:

---

is the result likely to be better than the stock de-essers I have access to in Cubase

---

Depends on method and personal taste, but remember there are a lot of great de-essing plugins out there now that weren't available when this was written, with all the benefits of recall etc. I see Fabfilter have recently joined the fray alongside the likes of Sonnox, Waves, et al.

Quote alexis:

---

1) In the processing channels send cascade, would the 1st slot be the group that routes back to the sidechain, and the 2nd slot be the comp itself?

---

Different DAWS route in different ways. The input to the processing channel is a duplicate of the vocal track. The output of the processing channel goes to a spare bus which goes nowhere else. The compressor in the processing channel needs a sidechain input, and the side chain source needs to be selected to the processing channel's output bus.

Quote:

---

2) I've never "linked" two channels with a comp before, though I've seen that switch on my Fairchild - I'm thinking each channel would send to the Fairchild, where they are then linked ... but if that is wrong, I'd be grateful for any correction please!

---

Again, depends on the hardware/software compressor you're using. But in general, compressors that have a LINK option also have some means of deciding what other compressors are linked. In the most versatile that is sometimes via dedicated comp link buses. In simpler systems they just link to the compresor (of the same type) in the adjacent channel to the right. You need to find out how your compressors work, and just make sure that the vocal channel's compressor is linked to the processing channel's compressor.

You also need to make sure that the threshold and/or ratio in the vocal channel's compressor are set so that it doesn't do anything on its own (ie, very high threshold and very low ratio), but that it is switched on so that it reacts to the link control signals from the other compressor.

Quote:

---

3) Finally - Hugh, can you expand on the added flexibility this might bring compared to highly eq'd side chain signal in the comp on the main signal path you mention

---

Most compressors with side chain EQ have pretty basic EQ facilities. This technique brings to bear the full four or five-band EQ available in the console channel. It's also very easy to audition how you are adjusting that EQ, which isn't always the case with simple compressors. And you can adjust the threshold and EQ settings dynamically if necessary...

Quote:

---

... and (bottom line?) is the result likely to be better than the stock de-essers I have access to in Cubase?

---

Depends on your stock de-essers, obviously. Different de-essers work in different ways. Simple ones have simple HF boost in the side chain of a broadband compressor. More elaborate ones employ narrow-band or multi-band compression to avoid damaging the rest of the spectrum while de-essing. Some alternative designs detect the sibilance frequencies and actually subtract them from the signal...

De-essing is always a turd-polishing technique to some extent, and it's always better to deal with the problem at source rather than to attempt to fix it in a mix. However, the technology has developed over the years and broadband compression techniques are pretty old hat these days. There are more delicate hammers to replace the sledge!

hugh

--------------------
Technical Editor, Sound On Sound

Thank you Hugh, and BJG145!

It is the time for me to sit down and do the deed (well not literally the time, as I am away from DAW till after holidays). The SSSSibilance isn't too bad in the raw track, but I've got some parallel compression blended in with the dry vox, and that is where it seems to emanate.

Also, after doing some more reading, I realized that perhaps the reason my "Stock Cubase De-Essers" aren't working so well is that I am applying them after an exciter in the chain, which may be adding too many harmonics for the de-esser to deal with well. I will give the de-esser another chance by applying it just upstream of the exciter.

Thanks again!

--------------------
Alexis -Cubase 6.5.0/SX3.1.1.944, XP SP2, 4GB RAM (1GB not accessible, but used just to balance the computer so it doesn't tip over); Delta 66 in Omni i/O Studio; Motif8; UAD-1

(Incidentally there's a more general thread on de-essing here which might give you some ideas. As for the order of processing to apply, it's an interesting point...I'd probably do the de-essing before adding compressors, exciters etc to the equation.)

I agree -- sort the problems first, before applying creative processsing. So de-ess the raw track, then compress.

Slightly concerned that the track is sibilant yet you still want to add fake harmonics from an aural exciter. Sounds like the voice has qualities you don't want and is missing those you do. That says 'need a different vocalist' to me!

H

--------------------
Technical Editor, Sound On Sound

Quote Hugh Robjohns:

---

That says 'need a different vocalist' to me!

---

Not a different mic?

--------------------
Ohm's Law states, "Your PA isn't as powerful as you think it is".

Yeah, maybe a different mic, or a different mic position.... but assuming such basic things were already optimised and the mix is about 'polishing', perhaps the vocalist is actually the core problem...

H

--------------------
Technical Editor, Sound On Sound