Even with transducers, it comes down to someone thinking, "Maybe this will sound better. Let's build it and see." There's a little bit of theory that goes into the the guesswork of "maybe this will work better because of my understanding of XXX principle", but that's still guesswork. This is also absolutely true for speaker enclosures as well. Sure, there are guiding principles, which are an amalgamation of the knowledge from previous trial and error, but you will still have to build and confirm the theory even when you're relying on those guiding principles.
*IF* it sounds better, then *sometimes* they work backwards and ask, "Why?", come up with a hypothesis, and then create an experiment to test their theory.
This is fairly rare, because there's very little benefit to going through this effort. That's for scientists, not engineers. From a business perspective, it's more important that something IS better, rather than knowing WHY it's better.
This is easy to understand with "cables and such", but the same principle applies to creating new high end capacitors, designing crossovers, choosing the parameters to drive your tube(s) at, which tubes you choose (or which opamps you choose), transformer materials, internal wiring designs, pots, power supply regulators, you name it.
Sometimes it's a cost:benefit analysis, "Does paying the extra for a higher quality component make a difference in this part of the design?"
It still just comes down to, "Let's build it and see."