gents, ponder this.
first i hate bad math, especially when used arbitrarily, and without
bit-clean tests to verify the process works without errors. i'm talking
about Normalization, Leveling and what have you.
to me the best normalization (i hate that word too) would have a
left-bitshift or right-bitshift followed by a an offset of (x) dB to
achieve -24LKFS, undithered. of course gain is done in a single multiply,
but conceptually, i like the left/right shift like Sonic used to have.
this speaks to the issue of why some normalization routines are different,
better, sound-wise and transparency-wise. we can't see the math DAW people
put into the product on most DAWs.
the proposal is this:
-24dBLKFS basically will completely honor the crest factor of the best
(interestingly so would setting the referenced +4 (0VU) to -20dBFS)
it would make all archives compatible with broadcast/netcast/youtube
(that's a wish)/... and remove intermediary steps for derivative access
copies. ya know, there's a volume control on every device. although i
understand the convenience to the client, without completely unattended
batch processing, you will wind up burning human time doing something that
will be louder than broadcast standards, softer than commercial CDs, and
sits in yet another amplitude universe. at least it would match the one
standard that is available as opposed to having varying levels for access
copies. of course i absolutely see Karl's point on changing derivative
access reference levels in some cases.
i ask "is it possible that we can just use A85 and stop there", have a
standard playback level, completely avoid limiting, reduce humantime and
eliminate differences between access copies.
to restate it, for years i've thought we should make a digital standard
that is *based* on analog tape. we did that to some degree
with +4=-20dBFS. The 1770 gives us a clear opportunity to make an archive
without any worry about peaks whatsoever, no need for subsequent derivative
alterations *except* going from the master captured file (9632 A85 wav for
example) down to an mp3 for the client.
linkedin profile <https://www.linkedin.com/in/seva-david-louis-ball-434700/>