What validation test and results would prove that DBT is not working?
It sounds to me as if this an important piece of information in order
to get this discussion any further.


That is easy. One can insert something that will generate a *known subtle
audible difference* into the random samples.

What's a *known subtle audible difference* that *hasn't* been determined
by controlled comparison methods?

Any examples you'd care to offer?

This seems like a strawman complaint. I never said that the "known subtle
audible difference" had to exist outside the thresholds of human hearing that
has been established in well documented scientific studies.