Re: GG: WTC II and SBM

["John P. Hill" <jphill@frank.mtsu.edu>]

Greetings to the Collective!

Wow...the list is buzzin' again...

Re:  Super Bit Mapping
SBM is Sony's version of what's also called redithering and/or
noise shaping.  Essentially, it is a method of mapping 20 bit
digital information on to the 16 bits needed for CD mastering and
replication.  So...the original GG analog masters are played
back, converted to digital (probably at 44.1 kHz. with 20-bit
encoding), further mastered (EQ, compression, etc.) in the
digital domain, then redithered down to 16 bits to prepare the
CD glass master, which eventually is used to stamp out the CDs
you buy.  There are other methods for doing this redithering,
such as the UV22 process from Apogee Electronics, but Sony's
version is called "Super Bit Mapping".  This allows the higher
resolution of the 20-bit A/D process (yes, we're talking
*amplitude* here) to give us 16 "better" bits of information
on to the CD.  With stock 16-bit A/D conversion, the actual amplitude
resolution is somewhat compromised by other data overhead requirements.
In addition, conversion technologies have improved vastly since
the introduction of the CD in the early 80s.  Hence, each time
the conversion process improves, we can do better masters, or
at least get a higher *quality* of 16 bits on to the CD master.

The CD, of course, is about to be made obsolete by DVD (digital
variable disc) which will likely give us 6 or 8 discrete channels
of digital information (full surround, not matrixed as in Dolby
ProLogic) with 96 kHz. sample rate (better high frequency bandwidth
and transient representation) and *24* bit amplitude resolution.
Instead of settling for some 65,000 or so amplitude levels, we will
have close to 17 million!  That's a BIG improvement in low level
resolution and distortion.  The "music-only" spec for DVD has not
been agreed upon yet by the big audio poo-bahs, but it's on the
way.  It's going to be a *major* step forward.  Many of us in professional
engineering and production have not been happy with stock 16-bit
digital audio (either in the recording or playback chain) for some time.
Analog is not dead and some have never abandoned it.

Sorry if this sounds like techno-garble.  For a good introduction
to digital audio, try the first few chapters of Ken Pohlmann's book:
"Principles of Digital Audio" (3rd ed.).  It segues quickly into
multivariable calculus (and worse), however, so be forewarned.  As
John Woram once said:  "any sufficiently advanced technology is
indistinguishable from black magic...".  How true.

Another issue here is that the original GG analog master tapes
aren't getting any *better* as time goes on.  In fact, they've
probably had quite a bit of wear and tear.  Oxide does shed
over time, with the resultant loss of high frequencies.  There
are so many variables involved in mastering and remastering that
wide deviations in sound quality are possible and often easy to
hear.

Re:  1966
It's interesting that you mentioned 1966.  This is around the
same time that the 2- and 3-part inventions were recorded and
I've noticed some similar clicking and distortion sounds on this
GG Edition CD.  I'm kind of surprised that they didn't "de-click"
these in mastering, as that's pretty easy to do on hard-disk based
editing systems like Sonic Solutions.  How the clicks got there is
another issue; I had the vinyl editions many years ago, but they
stayed in LA when I moved to Nashville.  I still love this recording
though, with all those nervous "ticks" in the midrange of the
piano.  We've discussed it before, but I have the feeling that
GG really *liked* this peculiar post-surgical sonic quirk that 318
brought to the sessions.

jh


On 4 Sep 1997, Paul Fawcett wrote:

> Hi all,
> 
>    I finally picked up a copy the other day of 2 CD set - making up the 2nd book
> of the WTC, after having owned WTC I for some time.  I was quite surprised by 
> the poor quality of the recording in some of the pieces on the first CD....there
> seemed to quite a lot more "crackling" type noises in the mix than other Gould 
> recordings I've heard.  It almost sounds like electronic glitches, rather than 
> background noise from the chair squeaking or whatever.  This seems to be 
> particularly a problem with the pieces recorded in 1966 (according to the liner 
> notes, WTC II was recorded in 8 sessions from 1966-1971 and was originally 
> released in three LPs)   The notes also mention that the session occuring in 
> 1966 may have been cut short.  Was there some problem with the original 
> recordings (perhaps from the first LP), or was this noise caused by 
> deterioration of the analogue masters before they were digitally remastered?  
> Has anyone actually owned/heard the original vinyl, and if so, does this noise 
> occur here too?  This has all got me wondering if Gould became increasingly 
> picky about what he chose to release as his career progressed (in terms of sonic
> quality, not content).  Was there a point in his career where pressure from the 
> label might cause him to release stuff that he felt to be less than his best? 
> For example, did he always use tape splices, even from the earliest days, or was
> this something that he started doing later? Perhaps he refined this technique 
> over the years and used it more and more?  I do see from this recording that 
> Gould would sometimes have an exceptional day ... I can't remember the date (I 
> don't have the CD here at the lab), but there was a particular day represented 
> on the 2nd CD that was just magic....everything he recorded that session is just
> exceptionally fine. Do you suppose this was because he especially liked the 
> pieces being recorded, or just because of natural fluctuations in technique (or 
> perhaps mgs of Librium consumed...can we check this?)?  I had always thought of 
> Gould as having very consistent technique, but perhaps a closer session vs. 
> session comparison might be revealing.
>     In an unrelated question, I was wondering if someone could explain to me the
> idea behind Sony's Super-Bit-Mapping.  I understand that the digital remastering
> was done using 20-bits, and that CDs use 16 bits, but what do all these 
> proprietary bit mapping schemes actually do?  It seems to me the problem of most
> accurately mimicing the master should be trivial:  say one is converting a 
> number between 1-20 to a number betweeen 1-10... In this case, original numbers 
> 1 or 2 would become a 1, numbers 3 and 4 become 2, etc.  This seems to be 
> absolutely straightforward, so I presume that I'm missing something, and that 
> bit-mapping schemes are more sophisticated than this.  Do they bias towards the 
> high or low end of the range (acting as a compressor)? Are the values assigned 
> dependent only on the value input, or does it depend on context (perhaps the 
> values of the numbers before and after)?  What does this really do to what we 
> hear?  I also presume that the record companies make their digital masters using
> 20-bits so that they won't have to go back and do it all again when the next 
> generation of CD-players comes out.  BTW, do the 16-bit numbers of a CD just 
> indicate the amplitude of the waveform at that point?  If so, how many 16-bit 
> numbers are required for 1 second of music?  Limited Gould content, I know, but 
> I have a feeling that he would know all about this stuff if he was still here.
> 
> 
> Cheers,
> 
> Paulito
> 
> Paul Fawcett                            .***.   .***.           .***.        
> Department of Genetics                * | | | * | | | *       * | | | * 
> University of Georgia               * | | | *   * | | | *   * | | | *   
> Athens, Ga                        * | | | *       * | | | * | | | *       
> fawcett@bscr.uga.edu                '***'           '***'   '***'          
> 
>