I was thinking about doing the same. Happy here are more people who are into this! hello
Just some thoughts:
psupine
I have read somewhere that the Really Nice Compressor is also digitally controlled analog circuitry made for cheap.
My point would not be the $ but the versatility, the best of two worlds, as the Distressor does. (I found a blurry pic of Distressor's inside: https://www.gearslutz.com/board/attachments/geekslutz-forum/31825d1172571155-gutz-distressor.jpg )
Imagine:
We could emulate the compression characteristics of any conventional compressor out there.
"It could run as a compressor, limiter, expander, ducker, or de-esser".
Nick Yes, and other dynamic processing like transient shaping. And effects like negative ratio or crazy modulations of settings, for example. And of course saving presets and morphing in between them.
Digitally controlled analog is already done with stuff like ES Silentway + Eurorack VCA ...
The + of Bela (or any microcontroller) would be its low latency and the independence of a laptop for working in realtime in a live or studio setup.
"Perhaps some of the undesirable characteristics of the analogue control device could be cancelled out by driving it with inverse transfer functions. This means for example that the particular character of an LDR / LED photocell, if used, could be cancelled out opening up the possibility to make it sound like a vari-mu or FET compressor just by changing the algorithm."
@Nick: Yes. Why would be a LED->LDR your first choice? I read, that they would not be linear at all?
But I think there are easy schematics for doing this. Being able to cancel out very nonlinear circuitry would make us able to do it even with more linear ones.
And the digital and analog sections would be galvanic isolated. (We can also use transformers for inductive isolation.)
Maybe a VCA would be better? Or it doesn't matter at all?
We are talking about compression characteristics, not colouring ...
For being able to make it neutral, i.e. to subtract the inherent behavior of its own, one would need to measure and interpolate the characteristic curves at different frequencies in the whole chain:
ADC-CALCULATION-DAC-LED-LDR
The CALCULATION-part is where the subtraction takes place and that is Bela's latency in the chain.
For being able to model the whole chain you would have to measure it from "the outside" ... with a linear mapping in the CALCULATION-part.
Please correct me if i'm wrong.
There are also DCAs (Digitally Controlled Amplifier) If they have a high resolution that would be probably a good way ... but even high resolution would bring some digital stepping ...
" The resolution and SNR etc of the digital side-chain only need be 16 bit with nothing fancier than Bela as this is outside the audio signal path. "
@Nick: Are you sure that 16 bit would be enough for a pro-audio unit, as the low dynamic resolution of digital is one thing why people prefer analog? I think for a dynamic processor it is important to have a huge dynamic range. This is actually the digital resolution that comes into the analog world here ...
In the input ADC one could do interpolations, but not so on the output.
Does Bela actually have pro-audio quality in mind or is it more a nerdy toy with lots of shortcomings for a high-end project?
Please correct me if im wrong.
My preferred setup would be the "best" of two worlds:
Very clean and transparent sounding analog components, we could modify a GSSL Compressor, for example.
Very fast and detailed digital sidechain, for being able to emulate lots of curve characteristics.
The more I think about it i am coming to this:
The sidechain made as an open system for being able to use it with different analog components and a proven strategy to calibrate the whole for making it neutral / linear (for being able to implement different characteristics after).
I still don't know:
- Would 16 bit really be enough?
- Is the noisefloor of Bela ok?
- Is it possible to attach better DCA's and ADC's with higher resolutions and less noise at the same latency?
How to flatten the stepping of digital? A small capacitor as filter?
- Would a self-calibration be possible and desired?
I am really liking the idea of using Bela for this, as it should be really easy to program everything in PD!
Even big part of the developing process could be done with any computer + simple compressor hardware, always with latency in mind. Once it works for one, then two setups, it would be easy to implement it in Bela with any analog compressor circuitry.
As Bela has a very stable latency(?) it would be even more exact, besides being faster.
This is a very interesting project!!!