Glenn Richards wrote:
Eiron wrote:

Someone has managed to reverse-engineer HDCD encoding:
http://www.srcf.ucam.org/~cjk32/hdcd

Yup, saw this - decent guy that wrote it, also sent me the source code
so I could build it on NetBSD (which I use for ripping). Compiled
cleanly and works great.

So now it's easy to decode an HDCD to a 24-bit file and see whether
each track has 'Peak Extend' and/or 'Low Level Range Extend'. Looking
at a few (King Crimson) CDs last night, only three had 'Peak Extend'
of which none actually needed it as they wouldn't have clipped
anyway.

The Peak Extend function works spectacularly well when CDs are properly
mastered. Without HDCD decoding:

http://intranet.squirrel-net.co.uk/images/itstime.png

ReplayGained to 89dB, signal off the CD would be normalised to maximum
no doubt.

Now with HDCD decoding enabled, again ReplayGained to standard 89dB
reference level:

http://intranet.squirrel-net.co.uk/i...stime-hdcd.png


That is a big difference, but why would they bother to do it that way?
Far easier just to master the whole thing 6dB lower - it certainly won't
be running out of dynamic range at the bottom end. I speak as someone
who owns an Arcam Alpha 9, an HDCD player (very ashamed of myself for
falling for it).

LAME can take a 24-bit WAV and dither it back down to 16-bit. The
results speak for themselves. And yes, the difference you can hear is as
spectacular as the difference you can see.


What do you mean?

d