> noise will be dominated by current noise which for a transformed Z of 5kohm will be about 40nV/root Hz at 1kHz and rising at lower frequencies.
Or say 6 microVolts across the audio band.
Note that the proposed operating level, ~1V, is about 100dB higher. When you really have 100mV-200mV at the mike, 1V after a step-up, your raw dynamic range is large compared to the recording/playback channel. When I'm set for orchestral peaks, direct to CD, I don't really care if mike-noise is 14dB SPL or 25dB SPL. True, there are better recording media (few playback rooms are even 96dB). But the loud sources that leap to mind are close-mike Fender and drums. S/N on Fender is just moot. Many (too many) drummers never stop, so noise is a non-issue.
Yes, a Darlington or Sziklai would give lower noise (and the Sziklai lower distortion).
But a nothing-special JFET would also give lower noise, would not need 100Ω series resistor (for short-protection if not for oscillation), and might be more euphonious, or look good on the product brochure. And no input DC current to sock the iron, and allow much higher bias resistors for smaller bias bypass cap.
In general I'd have to say most mike transformers are not meant to whack a volt or more output. I see Gus has pondered this.
If your mike levels are really this hot, many of the frills in mike transformers are pointless. You can get away with 20dB less shielding, quite a lot more parasitic resistance, etc.