After a short investigation: it's a normal behavior of OS, not aida's bug.
Last time I launched some test programs: dxdiag, then aida b2260, then b2267.
It was found that dxdiag set up some "core system" audio drivers regardless of actual audio devices present or not, and after closing dxdiag, these audio drivers hanging in memory (and shown in system devices) for a short time.
So the first report was with these drivers, but while second one made, they were unloaded.
Without dxdiag "disturbance" different build's reports in a part of "windows devices" are exactly the same.