quoted 7 lines At 12:37 AM 8/29/2003 -0500, you wrote:> At 12:37 AM 8/29/2003 -0500, you wrote:
> > > I know that vinyl is technically inferior to CD..
> >
> >actually CD is technically inferior to vinyl. check around the net to see some
> >of the tech specs.
>
> I prefer to say that CD is "differently inferior"
Not this again :-) Cut it out! This is a non-duscussion and it is very
old. Let me sum it up once again so everybody can stop cluttering
mailboxes:
What is the purpose of making musical recordings?
Your answer to this question is pivotal in this discussion. It generally
boils down to the following two choices: "So that people can listen back
to accurate reproductions of the recorded music at will" or "So that
people can listen back to cool music at will". I will get back to this
issue later.
Now, the nasties of the CD:
- Digital sampling at CD quality limits the frequency range of the
recorded signal to about 22050 Hz. This is well greater than the range of
human hearing, but it is nonetheless a limitation.
- Digital sampling introduces a 'quantisation error', which means that the
measured signal cannot be recorded exactly. The recording equipment has to
round off the measured values to the nearest predefined signal point, thus
producing a 'stairstep' effect, like on your monitor when you draw a
diagonal line (but less).
- Digital playback introduces 'jitter': small (picosecond range)
variations in playback speed, due to the imperfections of electronic clock
signal generators.
- Digital audio is limited in dynamic range because of the bitrate. At 16
bits, CD can offer at most 96 db of dynamic range. An orchestra needs
about 140db. Dance music fits well within the range.
The nasties of vinyl:
- Crosstalk. The needle of the turntable plays both the left channel- and
right channel data simultaneously. When a bump occurs on the left side of
the groove, the needle is pushed right and an electronic current results
in one of the stereo channels. When a bump occurs on the right side, the
reverse happens. Obviously, it is a pipe dream to think that bumps on the
left side will not be audible in the right channel. The crosstalk between
the two channels is usually as much as 40db, which is an incredible lot.
- Phase problems. From the above, you might wonder what happens if two
bumps occur on both sides of the groove at the same time. Answer: the
needle will pop out of the record and possibly wreck your turntable.
Because of this it is asolutely impossible to press signals with heavy
phase differences onto vinyl. Vinyl plants _always_ heavily preprocess the
audio before they cut it to remove these differences.
- Turntable motors are even more horrible at timing than digital clocks.
- Scratches, dust, the sun, wear, etc., etc.
Now comes the question: which one is 'best'? Well, that depends on your
answer to the earlier question. If you think a recording should be
accurate in a mathematical sense (*) the disturbances caused by the CD are
absolutely dwarfed by the problems of vinyl. The signal reproduced by the
CD is a lot closer to the original mathematically as is the vinyl
recording.
Now, things get hairy..
- The type of heavy preprocessing done on vinyl records is often
considered to sound 'fat'. But.. people have been doing it to CD's as well for
the last decade, because everybody seems to like it.
- While crosstalk is absolutely a bitch on vinyl, it _sounds_ less
disturbing to the ear as digital quantisation noise. But.. digital quantisation
noise has been greatly reduced by smoothing output filters and shaped dithering
since the earliest cd players. Besides, many CD players and digital sound cards do
oversampling ( a form of interpolation) on the outputs, taking the
quantisation noise well out of the audible frequency range.
- In light of all the processing and post-processing that is done in
studios, can you still speak of an 'accurate' recording?
- Your stereo and living room suck in comparison to the monitoring rooms
of studios. It will not ever sound like it did in the studio.
- Isn't the 'best' really only what you believe is the best? Don't get
all tribal over it.
The moral to this story is: don't waste your time on a discussion that has
no clear winners.
Cheers,
Denis
(*): you can mathematically analyze the difference between two audio
signals, because an audio signal is in fact a point in a mathematical
space (the Hilbert space). Each conceivable recording is a point (vector)
in that space and you can measure or calculate the distance between these
points, much as you'd calculate the distance between two points on a map
or in a three dimensional space. If you'd do this for CD and vinyl
recordings of audio signals, vinyl would be orders of magnitude further
away from the original than the CD, because of the forementioned
crosstalk (the crosstalk is much louder than the CD's quantisation noise).
Another way to look at this is through information theory: it can be
proven that a signal at 44100 Hz (CD quality) can 100% accurately
reproduce any signals that are band-limited to 22010 Hz, which completely
covers the range of human hearing (which is 20000 Hz). The problem of
quantisation noise arises because this result from information theory
assumes 'perfect' samples, whereas sampling hardware has to round off. So,
the only difference between a CD and 100% accurate reproduction is the
samll amount of roundoff at each sample. With increasing bitrates, the
amplitude of this signal becomes nanoscopic, whilst 40db crosstalk is
definitely macroscopic.
---------------------------------------------------------------------
To unsubscribe, e-mail: idm-unsubscribe@hyperreal.org
For additional commands, e-mail: idm-help@hyperreal.org