set vs otl, which is the most musical

Staying away from all the inevitable rehashes of the pitfalls of these two designs and, instead, simply focusing on the question at hand, I think the majority of listeners who have experienced the unique charms of a superior OTL and SET contender would conclude that the SET is slightly more in the camp of the beautiful/emotionally involving while the OTL leans more to the unvarnished truth/invisible conduit persuasion. Both are very compelling in their own ways. Ultimately, I opted for the truth. Continued exposure to the truth has lead me to the portal of musical beauty.

I agree with Acresverde. (I started to explain but determined that I can't say it better than Acres.)

I've owned only one pair of SETs, Antique Sound Lab Explorer 805s, that drove 86dB-(in)sensitive Quad 989s. As I don't listen at high levels, they and I matched very well until I tired of the 989s' treble balance and changed speakers.

But with a single output trode per channel, power output of an SET is always an issue unless one has very sensitive speakers. Mine are the opposite (at 84dB sensitivity), so I've opted for all trodes but lots of them, in the form of ASL Hurricanes in triode.

If I win the lottery, I'll buy a pair of 140-Watt Atma-Sphere MA-1s and keep one pair or the other. I love my ASL 'Canes, but maybe those OTLs will sound a little quicker, cleaner, bigger, etc.
.

Nice as SET's are, they are always a bit different, specially in the high frequency Area... I own OTL's from Atma-Sphere, and in my opinion, that's the Real Thing (or very close to it). They combine the best from all worlds, Body, a High Frequency Range to die for and a holographic speed which has to be heard to believe.
But, you need a matching speaker to get the best out of it ( 8 ohms+).

there is no way to generalize. set amps may be very linear in their performance or sound like a vintage tube amp of the 70's. otl amps may also have the characteristics of accuracy or "romance" . it is necessary to evaluate each amp on its own merites, regardless of its design.

generalizations are usually wrong, most of the time.