While I agree with your other points your claim raises a genuine question for me. Please refresh my memory. What is and where is your evidence that there is noise coming from the MEMS, and that Yuval has not removed this noise. If there is actual noise such as a hum, tone or hiss coming from the MEMS itself that is a problem I agree.
This leads to a decision tree: My points below may not be comprehensive but they are the most obvious ones I could think of off the top of my head.
1 Is there an audio sound artefact getting into the playing of the chip? Where is it coming from?
If there is an artefact present in the audio from the chip that can only come from a few places: a) the signal that goes to the chip, b) the chip adds it through its own operation, c) some other auxiliary device or sound generating process adds it in the course of the playing being performed. Solving these problems requires quite different kinds of solutions.
If the sound originates from the chip itself this is serious and would undoubtedly prevent the chip being taken forward to mass production before it was solved.
2 Is there an extraneous interference being picked up or recorded along with the generated sound?
This could easily come from equipment in the immediate environment of the apparatus. Solving this problem requires a separate solution. If it is known to only affects demos done in this particular lab bench setting this may not need solving at all.
3 Is the sound present in the actual original recording or only present on the recording that has been heard by watchers of the video on playback?
This may be totally unrelated to the chip, could come from anywhere, and again may not be necessary to "solve" at all.
What I find hard to accept is that the executives and tech developers of the chip would knowingly tolerate an artefact in the output from the chip, while submitting it for manufacture at a mass production level, unless they have total confidence that that artefact either has nothing to do with the chip itself or that a well proven and well accepted fix has already been incorporated into the design that will be manufactured. The implications of that not being done are very bad, so much so that I cannot imagine either AP sending it to EM, or that EM would accept such a commission because they would know that the product they mass produce would never sell and would thus never provide any revenue for either AKP to pay their bill to EM, or for EM to market usable audio products into China.
So again I am eager to know what you are seeing as or claiming as direct and incontrovertible evidence that the chip itself produces undesirable noise, and that this has not been dealt with.
(20min delay)