Phil Leigh;289459 Wrote: 
> I'm thinking it's that the error was not a rounding error in the 24th
> bit, but actually "rounding errors" happening much higher up....say at
> the 16th bit or even higher...that was caused by the maths involved in
> the volume control stepping.

Sorry Phil - I never responded to this.

Certainly it's possible that there was at some point a bug that
prevented the firmware from rounding properly, and if so the error
could be arbitrarily large.  But if you round correctly (and we're just
talking about multiplying two numbers, it's not rocket science) the
error due to the rounding is always in the LSB of the bit depth you're
using.  

The procedure slimserver was using before the "bug" fix - at least as
described in the bug report I linked to - is proper rounding when
multiplying by a gain factor that isn't a divisor of 256.  What had
changed was that the new volume curve meant that -none- of the gain
factors (other than max, of course) were divisors of 256, so at any
volume setting there was a rounding error.  But as we've seen, that
error is totally negligible.  After the "bug" fix all of the gain
factors down to -30dB or something divided 256, so there was no longer
any roundoff error, but the volume scale wasn't quite linear in dB any
more.  Neither of those is audible, so it doesn't matter - the whole
thing was a waste of time.


-- 
opaqueice
------------------------------------------------------------------------
opaqueice's Profile: http://forums.slimdevices.com/member.php?userid=4234
View this thread: http://forums.slimdevices.com/showthread.php?t=45736

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/audiophiles

Reply via email to