I did a similar check on cables of various lengths during my undergrad. Basically just took some recording then plugged the output back into the input using cables from just a couple inches long (intended for connecting guitar pedals sitting next to each other) up to around 16 foot (I was a student and had no intention of spending money on 100 foot cables I’d never use again), also various qualities at the low end.

There was no measurable difference between any of them, whether it was a real audio signal being analyzed or even various types of noise (that tend to load up all of the frequencies).

The only difference between the different tests was the usual randomness of the lowest order bits on the recording device itself, which I believe is influenced by whatever EM activity is in the area, including the CMB, which can’t be cut out without cutting out the input signal itself unless maybe if you do the test inside a faraday cage.

And if you’re thinking “oh so then there might be a difference if you cut that randomness out”, yes, but anything you listen to is still more affected by that interference than any cable choice and you probably never noticed that interference in the first place.