emrr wrote:I'm tending to agree with the point the the software designers generally are not hardware designers, and tend to have less experience with what hardware actually does, and why. So they model the surface aspects, a few of the most obvious traits they get from some testing, and can't see in any further than that. All one has to do is look at the built in dynamics models in the AU system, or many DAW's. Many examples of control descriptives that have never existed in pro audio! Names created by someone from a coding perspective, who has no knowledge of industry language and standards.
I dislike stereotypes but I encountered a certain amount of that during 15 years working at Peavey where I was strictly an analog guy and often had to rely upon digital specialists from the digital group. I recall one case where a digital implementation of a dynamics function (gate-downward expander), that remarkably sounded better (smoother) than any analog gate I ever designed. I asked but never got a good understanding for how and why, and this was just one one minor function inside a cheap multi-efx unit, long obsolete now. Knowing the coder involved, I consider the fact that it sounded so good, pretty much an accident and not reflective of the software coder's chops.
I am aware of at a least a small handful of old school analog pukes who evolved into DSP. One notable engineer is Steve Dove who I have faith will get it right. Of course for every Steve Dove writing DSP code there will be 100 wet behind the ears coders without a clue. Not only are they clue-less but inconsistent. I have been ranting for almost a decade now that we don't even have a clear definition for Q in a boost/cut EQ, but that was true for analog circuits too. It's only more visible with digital control of EQs, where there is an expectation of accuracy or consistency to accompany the control precision.
In the area of loudspeaker processing there is a small handful at the cutting edge who not only work in the digital domain but have developed semi-custom platforms to support their advanced work. This is a case of a sharp engineer defining a task, and then creating the tool to accomplish that task. Digital is a very powerful tool.
Getting back on topic (sort of) dynamics processing strikes me as a rich opportunity for non-linear (digital) decision making. Much of the strategy behind dynamics processing is to conceal artifacts from short term gain changes, while realizing the benefit of the longer term gain manipulation. There are many individual techniques to accomplish this but they often involve trade-offs in complexity. In the digital domain multiple strategies can be overlaid using the same base hardware, and selectively applied using situational decision making.
Analog hardware and software (on digital hardware) are both means to accomplish an end. Many of the early classic or legacy design are mostly accidents or victims of how the gain change was effected. Some like the combination of luminescent panel with photo-resistor sound very good despite the simplicity of the approach, others not so sweet.
Digital hardware finally separates the performance of the gain element from the overall path integrity, but just like before we can not completely get away from the artifacts caused by raw gain manipulations. So the task to make it sound good has just begun.
I speak about what could be, not what is (?)... IMO digital dynamics "could" be superior... time will tell where we end up.
JR
Cancel the "cancel culture", do not support mob hatred.