)
Dallas Legan wrote:
>
> Regarding the after meeting discussion of DVD technology,
> I just checked and it the remastered audio CDs of originally analog
> material I mentioned are 20-bit, as I thought, not 24-bit.
> The difference from earlier, (I assume) 16-bit masterings are spectacular.
See the March 1998 "Dear Mr. Know-It-All" at:
http://www.scoug.com/os24u/1998/scoug803.2.mrkia.html
I remember the DVD discussion being separate from the audio one. I was
tossing out some old catalogs yesterday afternoon and two of them were
from 24-bit audio companies. Sorry, don't remember the company names.
By dithering the least significant bit (LSB), you get the equivalent of
a "half bit" value. Thus, by doing so your recordings are "less hard"
(a non-technical term). A 4-bit example: If you have the exact values
1100 and 1101, record them that way. But if you have a value that's
half way in between, record it as alternating 1100 and 1101 and you have
an in-between value. That's what the dithering does, and since you're
dithering at the sample rate (96, 48 or 44.1 khz) the dither itself
isn't supposed to be audible.
Dithers, however, are audible. A random dither is audible by itself
because it contains lower-frequency components (do a Fourier expansion
and you'll see them). An alternating dither (0, 1, 0, 1, etc.) is
audible (as are all forms of dither) because of its interaction with the
random signal being dithered (again, a Fourier will show the
subharmonics). The audible harmonics from dithering are generally more
pleasing than the hardness resulting from non-dithered recording, just
as the harmonics resulting from Class A tube amplifier overload ("even"
harmonics) are generally more pleasing than the harmonics resulting from
semiconductor amplifier overload ("odd" harmonics). (Some advertising
claims to have "new kinds" of dither which can't be heard. This is
impossible; however, some kinds of dither may be less noticeable than
others when used with certain kinds of audio material.)
My oscilloscope tests in the studios found instantaneous audio peaks
(i.e., occasional single cycles) 19 dB above the maximum VU meter
readings (a damped rms measurement method). I found phonograph records
with peaks between 13 dB (on a Doors album) and 6 dB (Elvis Presley's
"Blue Christmas" 45, an extremely hot mastering apparently for jukebox
play). Since each bit represents approximately 6 dB (it's not exactly 6
dB, do the math; log 2 == 0.30103) your 16 bit audio channel with 96 dB
of headroom (6 dB * 16 bits) is reduced by the required headroom (19 dB,
13 dB, 6 dB, pick a value) and you're somewhere around 80 dB. That's
pretty good, but at comfortably loud listening levels of 90 dB spl and
with human hearing audibility down to -10 dB spl you've got 20 dB of
noise (dither, distortion, ringing, whatever) that you can hear but
don't want to (the actual studio noises, such as the clothing noise and
breathing of the musicians, isn't as displeasing). Since 20 dB is
approximately 3 bits (6 dB * 3 bits), an increase from 16 bits to 20
bits results in a very noticeable improvement in sound quality.
There's more: When you're mastering a session with real musicians and a
real singer, you have to give yourself _plenty_ of additional headroom.
The musicians hit a hot passage and their levels go up 5 or 10 dB above
the levels you set during the rundowns; the singer leans into the
microphone and the vocal mic levels go up similarly. To keep from
clipping, you set your levels low, and this puts you closer to the
noise. 20 bits is no longer sufficient, hence the need for 24 bits.
Once the audio is recorded and you can "see what you've got", the mix
can be at 20 bits because you don't have to allow for unknown levels.
This directly relates to programming because, when you're dealing with
audio signals, you have to allow for the bits which eventually will
affect the recording's quality.
- Peter Skye
=====================================================
To unsubscribe from this list, send an email message
to "steward@scoug.com". In the body of the message,
put the command "unsubscribe scoug-programming".
For problems, contact the list owner at
"rollin@scoug.com".
=====================================================
<< Previous Message <<
Return to [ 19 |
April |
1999 ]
The Southern California OS/2 User Group
P.O. Box 26904
Santa Ana, CA 92799-6904, USA
Copyright 2001 the Southern California OS/2 User Group. ALL RIGHTS
RESERVED.
SCOUG, Warp Expo West, and Warpfest are trademarks of the Southern California OS/2 User Group.
OS/2, Workplace Shell, and IBM are registered trademarks of International
Business Machines Corporation.
All other trademarks remain the property of their respective owners.
