A few comments on the "transformers are the answer" topic. I agree that especially when you have an impedance mismatch, particularly the wrong way (high into low), that needs **transformation** rather than gain or attenuation (fractional gain),a transformer could be very good.
Its that "could be" part. They have many issues - several notes above, several not. And like most problems, they can be mostly solved with money, lots of money, and few have attempted it. No market maybe, or very very hard. An ideal, variable, volume-controlling transformer must be VERY expensive.
Ditto resistors. They have issues. You can buy better ones. But.... I actually think even modestly good (name brand 1% metal film) resistors are pretty transparent. And with resistors there is a mature market for VERY good ones, and while they too are expensive (like 10-50X the cost of modestly good ones), even at 10-50X they are still less than half a buck (Caddock, DALE mil-spec noise controlled, non-inductively wound, blah blah) or were the last i checked which i admit was years ago. To be clear, you need a dozen to hundreds of pairs depending on how you implement the solution. I use 11 pair. Oh, and a micro-controller to control the mess.
But look a the big picture: we are now worrying about two more resistors in the signal path, one series one shunt. There are likely dozens in your signal path already. Are you replacing all of those too? Or removing them and inserting.... what?
I firmly believe that for 99% of applications, a true, discrete, stepped attenuator using a minimal number of resistors, great contacts, and good (not awesome) resistors is pretty darn transparent. I can say for sure its revelatory compared to what nearly all of us have heard - and I can do A-B-C side by side tests since i have them all, independent of the preamp circuitry, with a switch to do the rest evaluations.
