Now you have to re-encode the MQA to target 20 bits, not 24.
Let's do that then... oh wait.. for that, we'll have to pass the hat around and collect vast amounts of money, just to get a few files encoded...
Am I the only one that sees that as problematic in itself? Don't get me wrong here, I would also like to chip in with $20 for this exercise. Still find it staggering that this is really the only way to get a bunch of files MQA encoded.
But back to the 20-bit issue. I don't really think it will matter a lot if you feed the MQA decoder a 20-bit or 24-bit file. It will use the lower bits (of the 24-bit output) for the ultrasonic content, so it would chop off some of the 24 bits anyway, it should just chop off less for the 20-bit file. The main point is the file size here. It really makes the whole MQA ordeal unnecessary if they end up being so close. So this definitely is something to consider adding to the bunch of test files for testing.
Is it already clear what test files will get encoded? What premises will be tested? And how will they be tested? I think it would be good to have these things sorted upfront so that it is clear to everybody what to expect.