If the highs are on the bright side, that could be a sign of distortion. Our human ears are more sensitive to distortion in the higher frequencies compared to the lower.
If it is distortion, there could be many reasons for this. The most common reason is lack of negative feedback from the output back to the input. Negative feedback corrects for the non-linear behavior of the output filters and if done properly, also improves the signal-to-noise-ratio. Ultimately stability and distortion are improved.
It's actually very challenging to build a high power Class D amplifier for a frequency-dependent load like a loudspeaker. Although I've never seen a magnitude and phase plot for the Model T, I would expect the impedance to vary from 2 Ohms to maybe over 30 Ohms and from an inductive load at lower frequencies to a capacitive load at higher frequencies.
If you ever find a plot, I'd like to see it.
