[net.audio] Significant Digit and Realism

shauns (11/08/82)

Whew! rabbit!jj seems to be really angry about the audio cowboying on the net.

I fully agree with his opinions of digital-it is definitely superior to analog,
independent of weighting curve.  However, I believe the point being discussed
was the ARTISTIC superiority of available digitally mastered disks, not their
sonic superiority.  At least, that's what I commented on.

rabbit!ark (who is rabbit, anyway?  Where is rabbit?) has brought up an interesting topic - distortion mechanisms in digital recording.

There are some recording studios, Sheffield Labs most vocally, that eschew
digital on the claim that its mid-to-high frequency definition and `airyness'
(sp?) are inferior to analog.  This claim seems somewhat suspect.
dbx claims that analog mastering decks when used with the dbx system have a
dynamic range of 120dB.  Also, a sidebar article in a recent edition of
Stereo Review enumerated some of the potential problems with digital, to wit:

=the behavior of THD in the digital format is exactly opposite that of analog-
 THD increases with decreasing signal level because of the finite size of the
 LSB.  However, for a 16 bit system this means that the signal would have to
 approach -60dB for a 3% distortion level.

=0dB is 0dB on a digital system-period. There is NO headroom, or at the most
 perhaps 5dB if you accept 3% THD as the definition of MRL.  This could be the
 cause of a lot of listening headaches.  The digital recorder doesn't
 `round over' the peaks like an analog machine but hard limits them.  This
 could throw a lot of HF(10-20KHz) Distortion products into the output-anyone
 who's listened to a tweeter breaking up knows the headaches it causes. But
 this is a problem of miking and mixing technique.

=The extremely steep filter slopes used to keep the sampling rate close to
 Nyquist induce phase aberrations in the audio passband.  If you use a crummy
 filter, this might be a problem.  However, it seems to me that phase jitter
 between channels is more of a problem-and analog recorders have it much worse
 than digital.  Secondly, can you actually hear 5 degrees of jitter?
 Perhaps the filter slope, if not steep enough, can introduce IM distortion-
 but is it severe enough to be heard?

=Excessive code dropouts may be another form of irritation that is easily
 perceptible (speculation only!).

 As for `only' a 20KHz bandwidth, that's fine with me. It's dead flat out to
 20KHz regardless of recording level-show me a home deck that can do that at
 MRL!

 One final thing-has anybody considered the truly marvelous universality of the
 DAD? ! Assuming a player that can take the vibration
 levels and is small enough, the format is sized to go from your home to your
 portable to your car.  No more tapes eaten by vicious car decks!

 Riding the dynamic range, the old cowpoke,

 Shaun Simpkins

 tekcad!shauns

nrf (11/09/82)

Re: distortion at low levels with digital techniques:  such distortion can be
overcome by using an alternate coding scheme from straight PCM, such as Adaptive
Delta Modulation, Differential PCM, Adaptive Differential PCM, or some representation
with a exponent-mantissa format similiar to 'floating point'.

N. R. Fildes, BTL Whippany

tony (11/10/82)

#R:tekcad:-25700:pur-ee:12000002:000:264
pur-ee!tony    Nov  9 19:02:00 1982


By the way, my recent note which gave some theoretical SNR
info related to quantization assumed PCM was used. If delta
modulation, etc. is used, then better SNR figures can be
achieved, REGARDLESS of the amplitude of the signal.

				Tony Andrews
				pur-ee!tony