[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: GG: CBC Recordings -- the best possible?
On Sat, 18 Jul 1998, Tim Conway wrote:
> I know that there are a number of expert sound-recordists (I hope that's
> the right term) in the list. I should be very grateful if they could
> answer the following queries I have:
>
> (1) What exactly do CEDAR and NoNOISE do?
> (2) Why are they not perfect?
> (3) Are there better processes for restoring original sounds?
> (4) What are the main problems facing someone trying to restore a sound?
> (5) What are the extraneous noises (I'm assuming one or more of the
> experts also has one or more of the CBC CDs)?
> (6) Why cannot the extraneous noises be removed? I would have thought
> that today's computer hardware and software would be able to meet the
> challenge.
> (7) Why is what I have called, in my ignorance, the wobble/wow/flutter
> still there?
> (8) Finally, is the answer to all my naive questions simply that if you
> restore too much you end up with something that sounds if not like ELO at
> their funkiest at least something alien to the GG sound?
Whoa....there's a lecture or two. Here goes, Tim:
1. CEDAR and NoNOISE are digital algorithms that work in conjunction with
hard-disk based editing systems like SADIE (in the case of CEDAR) and
SONIC (NoNoise). These systems are digital "workstations" that are used
for editing and processing music signals in a variety of ways. Both CEDAR
and NoNOISE use intelligent algorithms to lower or eliminate background
noise from older recordings. The noise being removed comes mostly from
analog tape that was recorded at lower tape speeds (7.5 or 15 ips)
without noise reduction systems (like Dolby A or SR). These modules can
also do de-clicking to remove spurious clicks and pops in the recording,
such as might be produced when transfering from an old vinyl LP or 78.
2. They are not "perfect" because getting rid of noise and extraneous
sounds after-the-fact is a very tall order! Once noise and distortion
have been recorded onto a master tape, it's in there with the music and,
in some cases, more audibly so. It's a bit like putting too much Tabasco
sauce in your chili during the prepartion and then trying to take it out a
week later (without changing the basic flavor of the chili, n'est-ce pas?)
One of the great advances with digital technology is that tape noise
really stopped being an issue. There are tons of *other* ways you
can make bad recordings or add unwanted sounds, but it did get easier to
have a clean recording at the end of the day without oceans of analog tape
noise.
3. No. Actually, the two systems you mention took *lots* of serious
research and development and are really pretty amazing at what they do.
The trick, of course, is to make a great recording and control the sources
of these problems on the front end. That's easier to do nowadays than it
was in the 50s and 60s at Columbia. The tools have improved a lot, as
have the expectations of serious listeners, like yourself.
4. That's a good question for someone working as an archivist. I'll
defer to any colleagues on the list for the moment on that one.
5. Control of extraneous noises and sound is one of the *most*
challenging aspects of acoustic, "classical" recording techniques. Ask
any engineer. Finding a good-sounding ambient space with low noise
("let's see if we can turn off all of the AC...") and immunity from
outside interference (planes, trucks, sirens, etc.) is a very challenging
task. Spaces that fill this bill tend to get booked *years* in advance.
Beyond *acoustical* interference, there's also the possibility (when you
least expect it) of *electrical* interference from induced 60 cycle hum,
radio frequencies (RF), etc. etc. Anyone who is willing to do a careful
audition of the '81 Goldbergs will be rewarded if, at various points, you
hear everything from subway noises to low-level radio station crosstalk in
some of the most exposed, quiet sections. The human ear and brain, BTW,
are *really* good at decoding all this low-level stuff that can (and does)
leak into your otherwise pristine recording! And, of course, these
problems often become apparent *after* the original session work, when
they are *much* harder to correct.
6. Well, we're back to the whole thing about "throwing the baby out with
the bathwater". It's better to nip these problems in the bud, before tape
gets rolled, but it's not necessarily *easy* or *cost-effective* to chase
all those things down. And, frankly, there were/are some really
incompetent folks out there making bad recordings (same with dentists,
lawyers, bricklayers, politicians, etc. etc.) CEDAR and NoNOISE can be
used on contemporary recordings, but they are more commonly employed on
archival projects.
7. Wow and flutter are time-based distortions that most often come from
mechanical imperfections in analog tape machine transports. These
problems do not exist with digital recorders, because the coding and
storage of information on tape is fundamentally different. Wow is
generally heard as low frequency "wobble" or modulation, whereas "flutter"
happens higher in the frequency spectrum. These distortions often are
periodic (cyclical), so they *are* predictable, but doing a reverse
pitch-shift algorithm to correct for this might (again) be a tall order in
the digital domain. I don't know of any modules out there to correct
specifically for this problem, so it ends up again becoming a part of the
recording (just like the sonic signature of the mics, the room and the
instruments being recorded).
8. Hopefully, what you end up with is the original *performance* in a
more listenable presentation. The goal of this type of processing is to
leave intact, as much as possible, the original artistic creation while
using all of the most current technology available to ameliorate the
limitations of older technology used in the original recordings. But the
analogies I made earlier with the chili and the baby/bathwater hold up.
Another way of putting it is......"there's no free lunch"....maybe just a
two-for-one coupon at the Subway.
Hope this helps,
jh