> I'm asking this in regard to the neg charge cap that controls the attack and release in a typical tube compressor. My first thought is I want the fastest charging type (if it exists) to get over that attack lag, but maybe I am asking the wrong question here, but still curious about the answer as is.
This is the WRONG question.
The situation you cite, tube compressors, attack time is not about the type of capacitor. Attack time is essentially the RESISTANCE times the CAPACITANCE. Resistance includes output impedance of sidechain amp (in simple limiters, often the ~~3K of the line-amp plates), ~~500 ohms in rectifier diodes, and any added resistance used to reduce attack time. In many simple compressors the rectifier is cap-coupled, which means it is a staircase generator, and will take many cycles to step-up (slower attack for bass, a good thing).
There is also under 1 ohm in the capacitor. However this does not increase the attack time, actually reduces it a hair.
You do NOT want infinitely short attack. It will "duck" on supersonic ticks. Even short attacks mung transients. Do 1mS-50uS, and clip the leftover hair.
If you do want "infinite fast" attack, transistor sidechains do it better.
Since you ask...
> I want the fastest charging type
Discharge is the same as charge.
Asking for the fastest discharge will get you in trouble (at least, if you know what numbers to ask for). The *fastest* caps are used to detonate the TNT around an atom bomb. At least one A-bomb project was detected when the plotters were asking a cap company for some extreme specs and would not talk about their application. I don't know, but suspect, that such caps don't have wires but BIG lugs like a car battery, to dump BIG charge REAL QUICK. (It's not so much the speed, but getting several discharges to happen at the same instant; if one TNT blows a bit before the others it just throws the atomic-stuff out the side and critical-mass doesn't happen.)
Dielectric absorption is a non-issue in tube compressors using any type of film-cap. The whole thing is approximate and loose-tolerance.
> an encode/decode circuit ...the encoder used a tantalum cap in the time constant circuit
Yes, every don't-care rule can be invalidated. Tracking codecs with a Tantalum time-cap is perhaps thoughtless. I suspect with tube-works it would be benign; with touchy little JFETs it could be an audible burble. (But far worse things have been put to/from vinyl.)
I don't recall a CBS vinyl NR? It wasn't to compete with consumer tape, surely?