HamCall (October 1991)/HamCall (Whitehall Publishing)(1991).bin/bcast/audvid/nutcra.txt

Text File | 1990-10-14 | 15KB | 263 lines

An All-Digital Nutcracker:
  Adapting Digital Technology for 3-Machine Mixing and Editing
By Bob Katz
Originally published in REP/July 87.  Edited and with additions for
Broadcast Forum Readers by Bob Katz August, 1987
---
  Conventional digital editing systems are normally configured to handle
a single master and slave VCR.  It represents something of a technical
challenge to modify such a system to accommodate a pair of replay VCRs
working with a master deck in a 3-machine mix and edit session.

---

  Caedmon Records is the oldest spoken word record company in the United
States.  Although Compact Discs are routine for Arabesque Records,
Caedmon's music division, they are still rare birds for spoken word
product.  That's why I was both pleased and surprised to hear executive
producer Ward Botsford's request:  Could digital equipment be adapted to
mix and edit a proposed Caedmon CD of "The Nutcracker Suite"?

  Botsford had digitally recorded Tchiakovsky's music in London,
masterfully performed by the Philharmonic Orchestra led by Michael
Tilson Thomas; in Caedmon's New York studio, he had digitally recorded
E. T. Hoffman's "Nutcracker Story", delightfully read by Christopher
Plummer into our M-S stereo microphone.

  Now he wanted to know if I could put together a system which would
control one VCR replaying the PCM-encoded voice track, another playing
the PCM-encoded music, mix the two together and even perform pickup edits
on the mix when necessary - all in the digital domain.  Suddenly I was
involved in an uncommon engineering challenge.  In fact, I soon learned
that this was to be a digital first: no one had ever taught the Sony
PCM-1610 and DAE-1100 editing system to dance like this.

  Before I was ready to say "Yes, we can do it," I inspected the video,
audio and digital patching of the digital equipment rented by Arabesque,
and made a cursory block diagram of a proposed modification.  I then
explained our proposal to Sony's field representative, who agreed that my
theories seemed sound.  But, until tested, there would be a degree of
uncertainty, since no one had ever tried this before.

  My idea was to use the DAE-1100 editor to the limit of its abilities.
Although the editor can control two playback VCRs, it can only start one
player automatically in preparation for an edit.  I could put the PCM
voice tape on this VCR, which would provide repeatability of entrance
(to the microsecond, as a matter of fact) as well as flexibility - we
would be able to "slip" the voice track against music.

  It was untenable to consider manually starting either the voice or the
music in a digital 3-machine mix situation.  First of all, there was the
unpredictable delay between starting a VCR and the onset of sound.  And,
more important, we planned to use the digital editor to perform pickup
edits during the mix.

  In order for the DAE-1100 or any other digital editor to perform an
edit, there must be an editing rehearsal, during which the operator
examines the digital bit stream before and after the edit point.  This
digital information is stored in internal RAM and used to perform a
crossfade at the actual edit location.

  It is absolutely necessary that the editing rehearsal exactly match the
real edit, or there can be a glitch at the edit point.  In our case, we would
have to ensure that the music and voice before the edit point were identical
both during the rehearsal and the real edit.

  The only solution, therefore, was to control the second VCR holding the
PCM-encoded music with a time code synchronizer.

  Ultimately, these turned out to be relatively trivial problems, but one
question would haunt us until the end:  Could we make the digital editor
cut into a composite audio mix, as opposed to the straight output of a
single VCR?

  The next step was to set a date for the mix/edit session, and arrange
for A.T./Scharff Rentals to deliver the equipment needed to supplement
the existing PCM-1610, DAE-1100 editor and two BVU-800 U-matic VCRs.  We
ordered an additional BVU-800, a PCM-1630 processor and a Time Line Lynx
synchronizer.  Caedmon's Neve console and EMT 245 digital reverb would be
used to mix and sweeten the voice and music material in the analog domain.

  With the technical arrangements under way, I sat down in Botsford's
office to hear about the project.  An analog version of The Nutcracker
had already been produced for the 1985 Christmas season.  The 3-record
(or 3-cassette) set was turning out to be one of Caedmon's best sellers,
with 45,000 copies sold as of January 1986.  Now the label wanted to make
a Compact Disc version, a deluxe 2-CD set.  Naturally, I asked him why
he hadn't thought about mastering the CD from the analog 1/4" (which would
also save 100 potential man-hours of 3-machine mixing and editing)?  He
answered my question by explaining that he wanted to avoid the accumulated
tape hiss from an analog voice tape and an analog music tape.  It would
be improper to dub that to a CD, he said.

  The first step in producing the all-digital Nutcracker would involve
re-editing the 1610-encoded originals of Christopher Plummer's voice,
conforming them to agree with the studio takes used to produce the original,
edited 1/4" master.  The next step would involve the music replayed from
Botsford's edited 1610 master of the score (and sold to CBS Records for a
music-only version).

  For the voice-over version, the music would have to be reorganized and
re-edited to suit the story, and preserve Botsford's original creative mix.
We mapped out the following schedule:

  --Monday and Tuesday: Ward Botsford would digitally edit the voice
master tapes during the day and in the evening I would patch in the
extra equipment and experiment until it worked.
  --Wednesday: Botsford and Katz would edit the music, using the original
    1/4" as a reference for where the music pauses occurred.
  --Thursday and Friday: We would mix.
  --The Weekend: Backup days.

  It turned out we would need all seven days and nights, plus the following
Monday to dub the CD master videocassette that was sent to Japan for Compact
Disk manufacture.

  On the appointed Monday, the equipment arrived.  I patched in the normal
DAE-1100 system and left for other chores.  When I returned at 5 pm, Botsford
reported that he had only managed about two hour's worth of voice editing,
what with constant interruptions by office business (does this sound all too
familiar?). Obviously, we would never get the project done that way, so he
promised to install a "Do Not Disturb - This Means You" sign on the door of
the control room for Tuesday.

  Monday evening was to the the ultimate test of my theories:  Could I make
the DAE-1100 edit the output from Caedmon's Neve console?  The equipment
certainly would accept a mix, for I could feed console tone into the 1610's
analog inputs.  When I tested the editor, however, my digital theories came
up all zeros:  the 1100 kept locking up, putting out nasty error codes in
little red letters.

  To say the least, I was disappointed.  Nevertheless, I spent the rest of
that evening studying the DAE-1100 and how it worked, pulling cables, trying
all kinds of cross patches.  Eventually I learned more about the internal
block diagram of the editor than the average field service engineer is
expected to know.

  Tuesday morning I gave Botsford the news.  I found a way to mix, but we
couldn't put together the pieces in the same pass, a very dangerous
proposition for such a long composition.  But I also told him that my Monday
night research revealed a possible solution:  All we would need was another
1610, bring the total to three units.  A call to A.T./Scharff revealed that
another rental unit was absolutely unavailable, but the firm's Josh Weinberg
volunteered to call Sony for us and see if we could arrange a loaner - after
all, no one had anticipated that we would need three digital audio
processors.

  While waiting for word from Sony, Botsford and I proceeded according to
schedule.  By 3 pm Wednesday afternoon, he had finished editing the voice
segments and called me in to edit the music submaster.  My plan on cutting
music was very simple: conform everything to the already mixed 1/4" tape,
duplicating as closely as possible all the original pauses.  Why fiddle
with what worked musically the first time around?

  While the music was being digitally edited, I replayed the 1/4" master via
the console's built-in cue speaker, manually varispeeding the tape for sync
with the music that was being digitally edited.

  All the music was laid down at unity gain, allowing us to ride level in a
subsequent mix.  In certain places where the music was to be faded out, we
gave the digital about 10 more seconds at unity gain and then faded.

  Laying down the music was not a very complicated process, except at certain
points in the work where the music ended or was faded and it was intended
for the voice to be heard alone.  For those occasions, I kept the 1/4"
reference tape (in cue) and the 1100 rolling - in effect editing silence
while listening to Christopher Plummer's voice in cue.  As soon as I
heard the music on the cue speaker, I punched a button on the 1100 that
marked the time code location at which the next piece of music would
begin, or fade up.  In this way, we constructed a music submaster
complete with silences, whose spacing and timing would match the ultimate
length of the two CDs (about 72 minutes each).

  Because we did not want to produce our submasters on U-matic C-75
videocassettes (danger of dropouts), we employed three C-60s of about
50 minutes each, counting overlaps required to cut them together later.
I knew there would be some minor timing error in this conforming process,
but made sure that the spaces would always be a little "long".  That way I
could correct any errors in the music track spacing by adding small pauses
(or room tone if necessary) to the voice segments.

  Later, during the mix, it turned out that the spacing was perfect 90% of
the time.  The other 10% was fixed by using the synchronizer's offset
function to tighten the music closer to the voice.

  By early Wednesday evening we had edited the music and, amazingly, were
back on schedule.

  However, our loaner 1610 was not yet confirmed, which meant we might not
be able to pickup-edit while mixing.  Botsford and I agreed it would be too
dangerous to stop the mix at each of the numerous voice entrances for, as
mentioned above, you cannot manually cue a VCR-based digital tape the same
as an analog 1/4".  We decided it would be better to produce a spaced
voice-only digital submaster, to be coordinated against the music master
we had just edited.

  Keeping the DAE-1100 patched for normal editing, I placed Botsford's
pre-edited voice segments on the playback VCR.  I then moved our music
submaster to the VCR that was slaved to the record VCR and set up the
console to monitor a mix of voice and music.  I also played the reference
1/4" via the cue speaker to determine what musical points Botsford had
chosen to hit the voice.  It took me the rest of Wednesday evening and
Thursday to digitally edit a properly-spaced voice tape (actually on three
C-60 U-matics) and produce a time code-based cue sheet for all the voice
entrances and exits.

3-machine mixing

  The good news came mid-Thursday: Sony would be able to loan us another
1610, scheduled to arrive Friday morning.  The other bit of good news was
that my spaced voice tape would almost let the piece mix itself.  In
theory, we would just have to start the machines and move faders for
two hours.  Friday morning we hooked up the third 1610 according to my
latest block diagram and attempted an edit.  To my relief, the edit worked
the first time.  We could mix two PCM-encoded tapes under programmable
control, as well as pick up anywhere in our mix and digitally edit.

  However, the processor playing back the music tape became intermittent,
and there were occasional glitches at the edit points.  By this time I
had developed a nose for digital problems and my nose was telling me that
the trouble was probably in timing.  Botsford let me have Friday night to
tidy up the situation, as long as I promised we would be mixing by Saturday
morning, come what may.

  That evening I found that the word sync (44.1kHz) patches I was using
were probably causing the glitches.  Changing over to composite sync and
routing the sync as shown in the block diagram of RE/P's July '87 issue)
proved to be the perfect solution.  Saturday and Sunday the equipment
performed without failure.

  The mix proceeded without a hitch; at one point we mixed for 55 minutes
without a stop or an edit.  In addition, the 3-machine mix approach was
well worth the trouble; it allowed us to slide tracks whenever necessary,
making the musical result even more satisfying.

  The sonic result?  Very pleasing...it was certainly the quietest
voice-over mix I've ever done.  If you ask me if I would do it again
I'd say: "Yes, with certain changes...but that's another story."


POSTSCRIPT AUGUST '87  for Broadcast Professional's Forum

   I would say that anyone attempting what I did with three digital pro-
cessors should first take a look at AMS's Audiofile, hard-disked based
digital editor.  An article looking at this intriguing beast appears
before mine in the July RE/P issue.

   By the way, a common editing problem (either with analog or digital)
occurs when cutting into a take made in a reverberant hall.  If the pro-
ducer forgot to have the performers play a little bit of the music that
preceeds the edit point, then the edit will not work because the hall
reverb will sound cut off.  Well, the method of editing I described in
my article, that is, editing a mix, can be easily adapted to fix this
editing problem.  Using two PCM-1630's, a DAE-1100 and only two VCR's,
you can send the output of the playback VCR into 1630 #1's D/A converter,
then into your console, where you mix it with artificial reverb.  The
output of your console (source mixed with reverb) goes into 1630 #1's
A/D converter and is patched into the Player A input of the DAE-1100.
Use the block diagram in July to check out the composite sync patches.
For the entire system must be in sync.  Using this method, you can
add reverb across a splice!

                -30-