> we don't really care about impedance at the high end.
It matters a lot, but gets complicated. The bass end is complicated enough.
> Impedance, in this case inductive reactance, is given by 6.28 f L
Or a worked example:
Say you have a core and winding that is 1 Henry. At 10Hz: 6.28 * 10 * 1H = 63 ohms; at 50Hz: 6.28 * 50 * 1H = 314 ohms; at 100Hz: 6.28 * 100 * 1H = 628 ohms; at 1KHz: 6.28 * 1K * 1H = 6,280 ohms, and so forth.
That is just the inductance. With real iron and copper you also have pure resistance, which acts in series with the inductance. Without defining "Q", just assume that the Q of an iron/copper inductor at 50Hz is generally no more than about 5 or 10. The inductive reactance of 1H at 50Hz is 314 ohms, so the resistance is probably 63 to 31 ohms. This is often called "DC Resistance", though it actually affects the low-end too.
So with no load on the secondary, the input impedance of a 1H iron-core transformer is 30 to 60 ohms from DC to about 5Hz or 10Hz, then rising to 314 ohms at 50Hz, 6K ohms at 1KHz.
This impedance "shorts" the source, so we have to ask what source impedance we will use. And as far as secondary voltage is concerned, the DC Resistance adds to the source impedance. Assume 30 ohms DCR. If we used a 315 ohm source, 345 ohms total source impedance, the response would be down -3dB at 55Hz and falling below that, but flat above that. Adequate voice quality; for music we might try a 100 ohm source, total source impedance now 130 ohms, -3dB at 22Hz. So we might call this a 100 ohm winding.
> You wouldn't want a 10K:10K transformer with a winding resistance of 1 ohm
You can't MAKE an iron-core inductance that you can call "10K ohms over the audio band" that has a 1 ohm resistance. One ohm of copper wire that fits on any core found in an audio studio won't have 10K worth of impedance until around 50KHz.
And usually the problem, especially in transformers that have to deliver power (output transformers), is that DCR wants to be too high. Like my headphone amp with 10K nominal load impedance and 4K ohms of DC resistance wasting most of the power. It is lucky (?) that you can usually manage to keep DCR down to 1/10th of working impedance far into the music-bass area.