["David Satz" <DSatz@msn.com>]

[Esther Gillie <esth@xxxxxxxxxxxxxxxxxx>]

From: "David Satz" <DSatz@xxxxxxx>
To: <ARSCLIST@xxxxxxxxxxxxxxxxxxxxxxxx>
References: <Pine.OSF.4.10.10102181933450.5035-100000@xxxxxxxxxxxxxxxxxxxx>
Subject: Re: arsclist Arhoolie foundation project.
Date: Sun, 18 Feb 2001 22:13:13 -0500
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
X-OriginalArrivalTime: 19 Feb 2001 03:12:14.0269 (UTC) 
FILETIME=[C1D146D0:01C09A21]

David Ackerman wrote:

> [ ... ] it is my understanding that the key difference between
> 24 and 16 bit is the decrease in the signal:error ratio that
> comes with the higher resolution. I would think that this
> characteristic of 24-bit would be most noticeable when dealing
> with low level audio.

This is a fairly common misunderstanding.  I don't think it's the
purpose of this list to argue technical details, so as a broad but
reasonably usable generalization:  given proper dither, the main
difference between 16- and 24-bit A/D conversion will be the noise
level of the result.  Low-level signals in a properly dithered system
are not a special problem.  The old scare stories about low-level
signals being turned into square waves and "digital deafness" at
low levels are holdovers from the early 1980s, when dither was
not always uniformly or properly employed in, for example, Sony's
PCM 1600-series studio recording system (which was used to
record the master tapes for many early digital recordings).

If that problem may be regarded as solved, then you need only
make certain that the medium you're using for a transfer is 10+
dB quieter than the source material, so that the process will not
raise the overall noise level.  16-bit PCM fulfills that requirement
by a very wide margin, given the source material mentioned here.

Not all 16-bit recorders have the best sound quality that can be
achieved in 16 bits, of course.  A particular 24-bit 96 kHz system
could very well sound better than a particular 44.1 kHz 16-bit
system--_or vice versa_.  Those specifications, without more,
simply aren't any guarantee of quality sound.  They can't be.


>> The only advantage to using the higher number 88.2, divisible by 2,
>> not 96K which either leaves more fractions when brought down to 44.1
>> or rules out using the cheap compact disc recorder at all, is that the
>> processing occurs with fewer fractional numbers before coming back
>> down to 44.1.
>
> I put the subject of higher sample rates on the table, while clearly
> indicating that this is an area of debate. Regarding 88.2 and 96k, yes
> the math is much cleaner to convert down to 44.1khz from 88.2khz.

The math may be somewhat simpler, but will you be implementing
the sampling rate converter yourself?  If not, then why is the ease or
difficulty of the algorithm your concern, rather than the result?  Both
hardware- and software-based converters are readily available which
can handle the conversion to 44.1 kHz without compromising sound
quality in any particular way.

However, I think for the purposes being discussed here, a good 16-bit,
44.1 kHz CD-R system is far more than adequate--there is no need to
deal with the problems of storing 24 bits or converting sample rates.

--best regards,

David Satz
Brooklyn, NY
Esther Gillie, Sound Recording Archivist		Phone: 716-274-1330
Eastman Audio Archive				Fax: 716-274-1088
Eastman School of Music,  Sibley Music Library	esth@xxxxxxxxxxxxxxxxxx
27 Gibbs Street, Rochester, NY 14604