Shaping signal using the dynamics of another one.

[stonerbox]

As the title says, I guess I'm looking for some sort of side chaining circuit, right?

[Digital Larry]

If you're just looking to do a volume envelope then you need an envelope follower for A and a variable gain element - voltage controlled amplifier (VCA) for B.

The envelope follower is usually a full-wave or half-wave rectifier followed by a low pass filter/smoothing capacitor.  If the response of the filter is too fast then you will get distortion out of the VCA, so there are some tradeoffs between responsiveness and distortion.

[merlinb]

The Engineer's Thumb would do this if you modified the side chain for your own purposes.
http://www.diystompboxes.com/smfforum/index.php?topic=97123.0
It's a textbook voltage controlled amplifier circuit (the control voltage needs to vary from about 2.5V (minimum gain) to 4.5V (maximum gain))

[~arph]

You can do this with a ne570/571 as well. Pretty simple actually. Look up the roctave divider.

[stonerbox]

Thanks guys! I'm not entirely sure how to wire signal "A" into the side chain and how it actually works in The Engineer's Thumb, but I'll look into it and try to figure it out somehow..

~arph I'm not having any luck finding (and let alone understanding) the Rocktave Divider schematic.

.. and yes my grammar is perfect! (Swedish)

[R.G.]

This is the process involved in a number of vocoders.

A filter bank splits up an incoming signal into many frequency bands. A similar filter bank splits up a microphone input.

The dynamics of the various microphone filter outputs are impressed on the corresponding filter banks on the input signal, so that where the voice is loud, signal is loud, if and only if there is any voice frequency content AND signal frequency content.

It's many bands of what you're proposing.

It's easy enough to do. You generate the envelope of the control signal, and route that to the gain control of a VCA running the controlled signal. As noted, the NE/SA571 does this easily enough, as do any number of compressors. Anything that takes an envelope and uses that to control a gain can be cobbled to do it.

[Mark Hammer]

Many gates and generic "dynamics processors" have a sidechain input, so that the gain is controlled by an external source, rather than the input signal as the control source.  Classic case in point,  the David Bowie tune "Let's Dance".  To add a punchier feel to the tune, producer Nile Rogers mixed all the brass down to a single channel, and fed the mixed-down brass to a gate.  Then he plugged his guitar into the "key (external) input" on the gate and used his signature strumming style to map guitar dynamics onto the brass section.  That's why the brass on the tune sounds like brass, as far as timbre goes, but not quite like a brass/horn section as far as feel goes.

A number of noise gates have send/receive loops.  If you plug the instrument you want to hear into the receive jack, and plug the source you want to use for controllingpurposes into the input jack, you'll be able to map the dynamics of the controlling signal onto the heard signal...to the extent that the specific gate circuitry permits.

[mistahead]

So, extrapoloating a bit, one could use a basic gate circuit (as much as there are any "basic" ones) plus a few noise generators and a little filtering to create a synth note (as I've not varied the pitch yet) with some guitar dynamics.

Again we have the issue of latching onto a note (from the guitar, fundamental?) and feeding that back into a pitch shifting circuit... however we can buffered split the input signal before the gate and squash the input right down and shape it up as we have no interest in the clean sound except to drive the gated mix shapes and the fundamental - all of our output audio is coming from a noise generator.

And we've come close to inventing the guitar synth again...

[~arph]

Thanks guys! I'm not entirely sure how to wire signal "A" into the side chain and how it actually works in The Engineer's Thumb, but I'll look into it and try to figure it out somehow..

~arph I'm no having any luck finding (and let alone understanding) the Rocktave Divider schematic.

.. and yes my grammar is perfect! (Swedish)

It uses the technique here:

https://app.box.com/s/4448c718e6a9df7e5346

Look at page 15. The dynamic restorer.

[merlinb]

Thanks guys! I'm not entirely sure how to wire signal "A" into the side chain and how it actually works in The Engineer's Thumb, but I'll look into it and try to figure it out somehow..

Basically like this. I scaled it a bit so now the control voltage varies from 4V to 4.5V, so a 1Vp-p signal would cause maximum modulation. Not proven, but I think it would work...

[stonerbox]

R.G. I'm sorry, I was a bit unclear on how the circuit is supposed to shapes the signal, no voice/pitch altering, only the dynamics. Then again, I'm not so sure I completely follow you??

Mark Hammer Fun facts about Bowie!  A compressor with side chain controlling the dynamics of the signal will probably do the trick.

mistahead "And we've come close to inventing the guitar synth again..." Haha, yeah! Synth guitars FTW!

~arph Will take a look at that one, thanks!

merlinb Making things super easy for me here, cheers!  

[R.G.]

R.G. I'm sorry, I was a bit unclear on how the circuit is supposed to shapes the signal, no voice/pitch altering, only the dynamics. Then again, I'm not so sure I completely follow you??

What you have described is a circuit that takes in two signals. One is the control signal, one is the controlled signal.

The control signal contributes only its loudness information; it is converted to an envelope signal.

The controlled signal's loudness is shaped by the envelope signal of the control signal by some means.

Consider this to be all neatly packaged up in a little box, all of what you're describing.

Take in a music signal, and run it through a filter box with one input and four outputs. The four outputs are split into frequency bands, like a speaker crossover might do, but more accurate. Call the outputs Bass, Mid1, Mid2, and Treble.

Take a second filter box, and run a voice signal into it, getting voice outputs in the same frequency bands.

Now take four of your dynamics shaping boxes. One has Bass from the music signal as one input and Bass from the voice input as the other input. It has output only when there is bass in the voice input to the whole mess. Do the same three more times.

Now mix all the dynamics shaping box outputs together.

The output of this whole thing has bass only where the voice input has bass; mid1 only where  the voice input has mid1, and so on. The voice input signal doesn't appear in the output at all, but the amount of loudness at each frequency band controls the loudness of the music signal in that same frequency band.

[stonerbox]

R.G. I'm sorry, I was a bit unclear on how the circuit is supposed to shapes the signal, no voice/pitch altering, only the dynamics. Then again, I'm not so sure I completely follow you??

What you have described is a circuit that takes in two signals. One is the control signal, one is the controlled signal.

The control signal contributes only its loudness information; it is converted to an envelope signal.

The controlled signal's loudness is shaped by the envelope signal of the control signal by some means.

Consider this to be all neatly packaged up in a little box, all of what you're describing.

Take in a music signal, and run it through a filter box with one input and four outputs. The four outputs are split into frequency bands, like a speaker crossover might do, but more accurate. Call the outputs Bass, Mid1, Mid2, and Treble.

Take a second filter box, and run a voice signal into it, getting voice outputs in the same frequency bands.

Now take four of your dynamics shaping boxes. One has Bass from the music signal as one input and Bass from the voice input as the other input. It has output only when there is bass in the voice input to the whole mess. Do the same three more times.

Now mix all the dynamics shaping box outputs together.

The output of this whole thing has bass only where the voice input has bass; mid1 only where  the voice input has mid1, and so on. The voice input signal doesn't appear in the output at all, but the amount of loudness at each frequency band controls the loudness of the music signal in that same frequency band.

Wow.. now that would be very interesting!! A lot more versatile and musical than just one controlled gate. THIS IS A GO!

[bluebunny]

This is the process involved in a number of vocoders.

Check out this simple vocoder design for some ideas.  It's modular, so you can easily use bits of it as appropriate to your requirements.  Look at "Analisis Circuit.gif" (sic) to see how the carrier (controlling signal) affects the program (controlled signal).  Always fancied building this, but never got round to it...

[stonerbox]

I'm a bit over my head here, but then again that's where the fun happens.

[moosapotamus]

This does exactly what you are looking for.

RA Penfold Fuzz Unit

~ Charlie