OMG, not another audiophile discussion.
I got into electronics when I was a very little kid specifically because of my love of music. I feel prey to the whole "audiophile" thing, and its endless pursuit of specsmanship, when I was in high school. Pre and main amplifier manufacturers one-upping each other every week. Quoting spec sheets was like trading baseball cards decades earlier. But I got into actual electronic design, which led me to taking physics courses, which led me to take a "Physics of Acoustics" course. Extremely specialized, focused exclusively on high fidelity sound reproduction across the entire signal path.
In this class we had access to the university's anechoic chamber and so every student brought in their best speakers for testing. One by one we ran the chamber on our speakers. It had a turntable in the middle that rotated the speaker relative to the fixed array of microphones so you could get a plot that looked a lot like an antenna beamwidth plot. And the equipment also measured distortion in addition to the usual frequency response, phase response, etc.
What we proved in that chamber was that speakers distort 2-5%. That's not a typo... those are integers to the left of the decimal point. Some of the "hot" speakers, such as from Bose and JBL, were close to 10% distortion. Then the instructor brought in his baby: A Klipschhorn (anyone here even heard of those?). He claimed it was the best he'd ever seen, or even heard of. And it distorted over 1%.
What an eye-opener. I'd been obsessing over hundredths of a percent of harmonic distortion, picking poly caps from specific manufacturers and even specific manufacturing lots (!!!), etc. Meanwhile, the physical transducers at the start and end of the signal path were distorting in whole percents! We researched phono cartridges, both moving magnet and moving coil (the best were still several percent), microphones (several percent), you name it. The best physical transducer we found were certain headphones that had ultra-low-mass drivers, they could get close to 1%.
Furthermore, all of the physical transducers were fraught with phase distortion and phase intermodulation (unavoidable unless you have a separate driver for every instrument, and impossible to avoid for things like pianos that have multiple individual resonators), nonlinear amplitude compression effects, etc. The list is endless.
That week in the anechoic chamber ended my obsession with specsmanship in audio. The electronics was already so good it didn't matter anymore. The wisdom of that old adage about spending half your budget on speakers, and the other half on everything else, suddenly crystalized into gospel truth. Except that I extended it to say "spend half your budget on your
physical transducers", to include the input side (mics, cartridges if you're into vinyl, etc.) as well as the output side (speakers, headphones).
The takeaway truth: The worst electronics is already better than the best transducers. There is no point in wasting money to shave another tenth of percent off the distortion figure, because your input signal will already be corrupted far past that point before you ever get your hands on it and whatever is converting those electrical signals into sound waves for your ears will corrupt it again by at least as much.
"But what about digital sources like CD's and DVD's?", I hear the audiophiles scream. Setting aside the corrupting effects of the brickwall filtering necessary to avoid aliasing, and the quantization errors no matter how much you oversample, unless you're listening to nothing but synthesizers that were jacked straight into the mixing console there was a microphone or some other physical transducer in the signal path - and your signal is already corrupted. That beautiful human voice? Microphone. That sweeping orchestra? Multiple microphones. If it was acoustic before it was recorded, the corruption is already baked in. Period. And that's ignoring the further damage from your speakers or headphones... unless you have some secret cochlear implants that are better than anything of which I'm aware (and I track that technology pretty closely), the transition from electrical to acoustic is an insurmountable enemy.
Seriously... save yourself and your sanity. You don't have to trust me, do some research on your own. Think about the WHOLE signal path, from the original musician playing their instrument in the recording studio to your speakers in your home. All the magic film capacitors and opto-attenuators in the world cannot undo the damage that happens along the whole signal path. Accept that, and start making rational choices.
Whew. With that rant out of the way, let's address this opto-attentuator topic. Unless it's a fully digital board, guess what professional, studio level mixing consoles use for each channel's master fader? Conductive plastic linear pots. I used to design analog mixing consoles and we did a LOT of research on faders. Sure you can run the signal through an analog multiplier and just use a control voltage coming off the fader (to avoid running the audio through the pot), which is basically what your opto-attenuator does in a passive way, but guess what? That control voltage coming off the fader is still a component of the output signal. "Multiplier" means all input parameters are present in the output. Noise, glitches, wiper degradation, etc. are all still going to be present. You haven't eliminated the effects of the fader, you've just moved them around.
Meanwhile, you've introduced two more "active" components into the signal path (not active in the sense of semiconductors or tubes, but active nonetheless in the sense that their behavior changes with some control input). Active components are never perfect. They're not perfectly linear, they never have infinite dynamic range, etc. Even passives have nonlinear parameters, and the photocells in those opto devices are no different.
Engineers in the audio profession addressed the problems associated with pots decades ago. Where they once used carbon composition pots, they now use conductive plastic (CP). True for rotary and linear pots. Carbon comp pots do indeed have lots of noise (which you can see with any oscilloscope) due to the grain structure inherent in their elements. Basically, the wiper "skips" across the microscopically rough surface of the element and you get glitches as a result. But CP's are microscopically "smooth", and their wipers don't glitch like that. CP's also wear more gracefully without developing the scratchy, grindy noises that old pots exhibit. And their cycle lifetimes are 10-100-1000X better too. Pots are about as linear and non-distorting as any component you're going to find.
The bottom line: Just pass the signal through a good CP pot. Optoisolation isn't going to offset the effects of having a garbage pot just because the audio isn't "passing through the pot". And since you will thus need a good pot anyway, just use it the way it's intended. If the varying input/output impedance is an issue you'll need to buffer it, but even with matched opto's used in pairs they're going to have different absolute impedances, likely even channel to channel within the same "passive attenuator"... how come nobody is bothered by that?!?
A last word... am I the only person bothered by the phrase "passive preamp"? Is such a device built using "straight wire with gain" fabricated from sintered unobtanium? I have yet to see a passive device provide amplification, but hey, I'm open-minded and willing to learn!
EDIT: OK, transformers can provide passive gain for certain definitions of "passive" (like ignoring power and only thinking about voltage or current). Can't wait to see someone Rube up a "variable audio transformer" as a "passive preamp". The mind reels....