BASH and expired patents

If the input is a digital stream to start, then Class-D is no different from a DSD DAC, but at a bigger power scale. As with anything else, the devil will be in the details, but just like DSD, if you push the sample rate up, the filters become trivial. Between the time BASH was created and now, MOSFET performance has greatly increased, new MOSFET technology is cost effective, FPGAs for digital control have vastly improved at a given cost point, and manufacturing costs of power electronics have come way down.

With BASH, I still need to reconstruct the audio signal at some level with a PWM style power conversion stage and then I need to ensure my class AB amplifier is able to reject that noise. Attempting to keep the voltage across the FET/BJT approximately the same may increase linearity, but that would be balanced against new issues. Primary I just see BASH improving efficiency, but without the potential long term sonic improvements possible with Class-D at least an all-digital Class-D.

Honest, I cannot agree with much of what you have said.
I don’t see digital as at all a step backward, and see 24/192 or high bit rate DSD as vastly superior to any analog format, tape or vinyl, but really we only need to compare to tape, as any vinyl is either an analog conversion from tape, or more far more likely for the last few decades a digital recording finally transferred to vinyl.

I can see one liking the euphonics of vinyl, which is what most people consider analog, especially the reduced channel separation and increased cross-talk, which in the absence of a well tuned room, may provide a level of ambience that is lacking from a digital stream (but which could be simulated). Even that low level background hiss can add a sense of realism. But for pure accuracy of recording, vinyl or analog tape just does not have the "chops" to compete. It simply is not even close.

For me, I do see the filter as somewhat trivial as when my PWM rate is pushed high and my pulses reasonably symmetrical, quite possible today, then anything that is not music is pushed way outside the audio band and can be filtered in the analog domain such that anything near the audio band is well below anything a human could detect, and the same would be true with near ultrasonics so you don’t have any weird speaker breakup modes. Class-D also has some inherent advantages with a good power supply design for reducing IM and multi-spectral modulation in real music.

Long term, digital signal processing will allow corrections to what comes out of a speaker that simply is not possible in purely the analog domain. That pretty much makes it a necessity towards what may be consider our holy grail.