Ceanothus

1) You'll get the best outcome by getting the Spotify bitstream to the headphones and decoding there. Is that possible - don't know. If you've decoded in the phone, then the rationale of Aptx is that it's a better codec than Bluetooth gives you for standard. But it's your ears that judge whether this matters in your use case. 2) Supports and reliably delivers in real world situations are very different things. Especially with congestion in the 2.4GHz band. 3) Not entirely. A fairer comparison would be between a FLAC encoding of the CD and Spotify.

Most of what Mr Hanson has to say is that he doesn't like expressing "noise per-root Hz". But the thing about noise is that how much of it you measure depends on the bandwidth you measure over. As anyone who's looked at noise spectra FFTs knows - the more bins you use (narrower measurement bandwidth), the lower the noise floor comes. 3dB each time you double the number of bins. This isn't anything unconventional as a google for "per root hertz" will tell you. So if a "16 bit noise floor" is coming out at -144dBfs then it's probably being analysed with a 64k point spectrum. Equally that'll be why the graphs include the "16 bit noise floor" as a reference since unless you know the analysis bandwidth is circa 1Hz the scale for noise is meaningless.