Erm, I'm familiar with ideas like 'signal reflection' in both practice and theory. I've developed and worked with systems that operate at frequencies from FM radio up to over 300 GHz. Some of it for the old NPL.
The cables used in most domestic cases are very short in terms of the wavelengths of the signals used for spdif. And also the cables *don't* have a uniform impedance at lower frequencies. Nor do the connectors on the boxes usually have the 'correct' wideband impedance. Nor, indeed, will the links inside the boxes.
If the 'reflections' are worse at lower frequencies they're better modelled and understood as added *lumped* impedances. Probably easier to tweak that way as well. But you'd first need a way to measure them accurately.
That said, if you think coax for audio is bad, you should check out the insane way HDMI is said to 'work'. Its a scunner.