From: Matthew P. OReilly <moreilly_at_moreilly.com>
Date: Wed, 26 Mar 2003 21:49:06 -0500

I think I see what you're saying. I was thinking in terms of linear
truncating of the least significant bits rather than whole-scope analysis of
each bit. Which is why an mp3 of a CD track sounds so much better than a
8-bit conversion of the same track uncompressed.

Ahhhhh.... the light goes on.



The only reason I do step-up bitrates is when a recording comes to me at a
lower bit depth and I want to use a better bit-depth effect. (Someone
records a vocal elsewhere and brings it to me to mix into a track, so I add
24-bit reverb/delay/etc.) Then it has to go back to 16-bit to go on a CD.
Those extra 8 bits make a noticeable difference in the finished quality.

Thanks for the replies, guys!

Matt

-----Original Message-----
From: owner-rockbox_at_cool.haxx.se [mailto:owner-rockbox_at_cool.haxx.se]On
Behalf Of Chris Holt
Sent: Wednesday, March 26, 2003 8:18 PM
To: rockbox_at_cool.haxx.se
Subject: RE: mp3-editor?

There is no "scheme" commonly known as a recompression scheme, so the
quality of such a scheme is irrelevant.
MP3, like JPG is a lossy compression scheme. Where JPG tries to discard
file information that is not easy to see, MP3 tries to discard file
information that is not easy to hear. Each generation of compression is
compressing a file with discarded information that cannot be retrieved
and so quality is lost. Even worse with JPG is that file sizes can
actually increase over the first generation of compression because
complex artifacts introduced in the initial compression have to be
analyzed and compressed. Of course MP3 filesize (for CBR files anyway)
is dependant only on length and bitrate so this is not an issue.

Lossless compression schemes, like Monkey for audio or PNG/TIFF for
images, don't suffer from multiple generations of extraction and
recompression, so your example of 16-24-16 would work out with no loss.
(but I see no potential for gain in doing this either)

Chris
Received on 2003-03-27