If I'm rendering stems or other .wav files that will again be worked on within my DAW, say for faux mastering or to be imported by another program or DAW I ensure I select Dither Master & Noise Shape Master as stacked processing does introduce artefact in the higher frequencies. I assume, because the lossy formats converted/unpacked to PCM/DSD etc. have filled gaps with zeros before truncation, that the problem is somewhat greater.
Now, I render at 24bit if I need to move things or process further, as would most folk but processing by streamers etc. may introduce more.
I usually upload 16 to streamers as a 24bit upload strains my bandwidth/allowance.
Dan Worral explains it nicely & explains why it's not a problem and is a problem depending on one's needs.

Please excuse the rude title.
Note...the little trick with -0.1dB is cool.

The concise explanation is at 12.43 in the video.
At 15.59 he plays with MP3 in the same processing.
Pity he didn't do WMA as that's germane to the website.
Perhaps someone with the time & nous will try the process on one.

For my purposes, that is recording other "real" instruments at 24bit, the basic rules for me are :Dither/noise shape any time in 24bit. Dither Noise shape ONCE for 16 bit.

Given the relative cost of storage versus the time & expense of conversion, the higher cost for the "audiophile" version no longer seems justifiable to me other than marketing of course.

Last edited by rayc; 04/21/22 07:28 PM.
Cheers
rayc
"What's so funny about peace, love & understanding?" - N.Lowe