[time-nuts] frequency (absolute) accuracy in sound recording/playback

Peter Gottlieb nerd at verizon.net
Tue May 8 02:17:10 UTC 2012


   Yes, the nice thing about that is it is so easy (if you prepare ahead
   of time).  Here is what we used:
   http://en.wikipedia.org/wiki/SMPTE_timecode
   Both have their uses.


   On 05/07/12, J. Forster<jfor at quikus.com> wrote:

   When I was doing some video production, we would first "Black Burst"
   the
   tape. This was done from end-to-end as the lattice.
   Then we assembled the segments onto the that tape. The inserted
   segments
   were always an integral number of frames.
   The source deck for the playback was slaved to the prerecorded black
   burst.
   The system was closed-loop and the start of every frame wasa not
   dependant
   on uS differences between two hour long times.
   -John
   ===============
   > We had time code to sync a number of separate A/V recorders so that
   during
   > editing you can cut from one to another seamlessly. I didn't
   calculate or
   > look at how tight the sync had to be. The mobile cams could be out
   there
   > for a while, maybe an hour or more, starting and stopping to change
   tapes
   > and batteries, but with the time code it just worked.
   > On 05/07/12, J. Forster wrote: A movie may be 7000 seconds, and you
   may
   > need a fairly stable timebase,
   > but every movie I've watched is made up of short ( One area where
   accuracy
   > is important is not because of pitch (nobody can
   >> hear 1ppm differences), but because of the need to synchronize sound
   >> from different sources, particularly with video or motion picture
   >> frames.
   >>
   >>
   >> 1000 seconds (20 minutes, give or take) with the sampler off by 1ppm
   >> will be 1 millisecond out of sync, which is probably hearable, and
   is
   >> 1/30th of a frame time. A 2 hour movie (about 7000 seconds) would be
   7
   >> ms out of sync.
   >>
   >> Yes, we're not looking at needing Cs accuracy, but 10-20 ppm
   probably
   >> isn't good enough. So you're pretty much not going to be able to use
   >> the $10 oscillator in a can.
   >>
   >> So maybe a decent Rb, which is good to 1E-9 without doing anything
   >> special, wouldn't be a bad thing.
   >>
   >>
   >> And yes, there's a whole art to synchronizing stuff which was
   recorded
   >> or filmed with incorrect sample rates, or ones that are
   "slightly
   >> off".
   >> It wasn't too long ago that "quartz lock" for a motion
   >> picture camera
   >> was something that was a "special order" from the camera
   >> rental house.
   >> I used to modify PC video cards for external clock input so I could
   >> adjust the refresh rate to match the camera speed (aka gen lock).
   >> There's a time nuts challenge... synchronizing something normally
   driven
   >> off a quartz oscillator (however crummy) to a mechanical device (the
   >> movie camera shutter).
   >>
   >> And given the creative hierarchy on a set, it's going to be you that
   >> adjusts to them, not vice versa.
   >>
   >>
   >> There are directors who (for whatever motivation) also don't want
   things
   >> like timebase correction used. Since I used to work for a physical
   >> effects company, I thought that these guys and gals who are hung up
   on
   >> the "purity of the process" were wonderful, since they
   >> typically wanted
   >> "real" special effects, not something composited in later
   by
   >> optical or
   >> computer techniques.
   >>
   >> There's a whole industry supplying 24/48 Hz refresh hardware, as
   >> well. Well.. there used to be.. I'm not in that business anymore,
   and I
   >> don't see credits for 24fps video as much, so they probably just
   paint
   >> the screen blue or green or put registration dots on it and comp in
   the
   >> images later. (Yes, I'm one of those people who watch all the
   obscure
   >> credits at the end for things like assistant hod carrier and such.)
   >>
   >> _______________________________________________
   >> time-nuts mailing list -- [1]time-nuts at febo.com
   >> To unsubscribe, go to
   >> [2]https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
   >> and follow the instructions there.
   >>
   >>
   >
   >
   >
   > _______________________________________________
   > time-nuts mailing list -- [3]time-nuts at febo.com
   > To unsubscribe, go to
   > [4]https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
   > and follow the instructions there.
   >

References

   1. mailto:time-nuts at febo.com
   2. https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
   3. mailto:time-nuts at febo.com
   4. https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts


More information about the time-nuts mailing list