Sean and Pubul57, if transparency is what you crave (lack of distortion, I'll get to that in a bit), then it is to your advantage to keep the impedance of the speaker as high as possible. This is true regardless of the amplifier technology, tube, SS or class D. IOW 4 ohms and less is not practical **if detail (transparency) and general musicality is your goal**. If instead you crave volume, lower impedance speakers will winnow that out of transistor amps.
This issue is so profound that the Speltz autoformer is beneficial to any amplifier that has to drive 4 ohms **it can make the impedance translation better than any solid state amplifier can**. IOW, transistor amplifiers will sound better driving 4 ohms through the ZERO than direct.
Figuring out what 'transparent' is was a serious issue that we had to face before we could even build anything. We solved it by going back to the microphones- then building the best mic preamps we could, then running them directly into the amplifiers and then auditioning both amps and speakers. It also required someone playing an acoustic instrument, so we could compare live vs. reproduced.
This allowed us to build our first reference, which we used to audition master tapes, and from there, LPs (CDs were just barely getting started...).
Ultimately transparency turned out to be "a lack of audible distortion, which would otherwise mask detail". The human ear/brain system perceives sound using a certain set of rules that all humans respond to the same way (taste is not an issue here). In this case, the rule is known as 'masking', wherein a louder sound will mask the presence of a quieter sound. So, low-level distortion will mask the presence of detail. Remove the distortion, and the detail is revealed. The distortions we are talking about here are at very low levels, yet our ears, via the masking rule, can easily hear the difference, and yes, you don't need a master tape to tell; the phenomena of 'not having heard that before' in a recording you know well is completely valid- you remove some low level distortion and voila!