An interesting question and thoughtful responses so far. I have a
slightly different take on the matter--give it a try.
I've done this for a slightly different situation, and found the
results satisfactory. First, you are correct, drift will definately
be a problem after a few minutes, and will obviously increase over
time.
In my case, I did a live concert recording for a friend--an outdoor
Miles Davis Tribute at Coit Tower in San Francisco. I ran a mini disc
off the main PA's board mix, then ran a stereo mic in the audience
onto a DAT machine.
I brought both stereo tracks back into pro tools and aligned their
start times to a convinient transient sound. I then measured the time
offset at the end of the program (over an hour). After some
head-scratching, I came up with the percentage difference in length
of the two files. Then, I samplerate converted one of the files to a
new value, the same percentage "difference" off of the nominal 44.1
rate.
Finally, I opened the new file and manually changed its sample rate
back to 44.1. This is just a header change, it doesn't re-sample the
file.
At that point, the two stereo tracks could be laid next to eachother
in the timeline and were the same length. And the alignment was very
good, as the correction was applied over such a lengthy period.
One might ask why I didn't to a time-stretch to match lengths; I
tried, but the result was too artifacty.
As I wrote this up, I realized an unintended concequence--my time
stretched track should also be very slightly pitch changed, but by
such a small ammount that I never noticed at the time.
In the end, I took these two files and made a stereo mix, using the
board tape as the main element and mixing in the audience mics to
taste. To enhance the sense of space, I actually purposefully delayed
the audience track another 15 or 20 milliseconds.
At some point, I indend to remix the project for quad, but that
hasn't happened yet.
So Lang, I'd give it a try. The only way to avoid the problem in the
first place is by locking both machines to the same clock, which
means one of them needs a word-clock input (or the ability to sync to
an external digital audio reference, rathern than its internal clock.
I'm unaware of any portable DAT with this feature, although I
wouldn't be suprised if the time-code HHB machines were capable of
it...
Rudy
>
>________________________________________________________________________
>________________________________________________________________________
>
>Message: 6
> Date: Wed, 06 Nov 2002 08:55:52 -0500
> From: Lang Elliott <>
>Subject: Re: Re: mosquitoes
>
>I have a question for all the tech heads:
>
>I want to experiment with some 4-channel recording using a special
>"quad-SASS" setup, however I do not have a four channel field recorder. In
>the short run, I'd like to try doing this using my two Sony TCD-D10 Pro DAT
>recorders, each having two channels.
>
>Has anyone tried this? If I record with both recorders, I could coordinate
>them by making a broadband "tick" at the beginning by tapping two coins
>together at a point equidistant from the two mike setups. Later, when the
>two stereo recordings are inputted into my computer (using direct digital
>S/PDIF transfer), I could slide the two stereo tracks to exactly overlap the
>waveforms of the introductory "tick". This should coordinate the four
>tracks, at least at the beginning of the recording.
>
>My question concerns drift.
>
>For instance, if I record for ten minutes straight, is it likely that the
>two recordings would drift in the time domain due to slight differences in
>sampling rates? If this happens, it could render the technique useless.
>
>How does a DAT recorder generate sampling rate? Would two DAT recoders of
>the same make and model have absolutely identical sampling rates, or would
>there be some inherent variation, however small? If so, then would this
>variation be enough to cause significant drift between the two recorders
>over a 5-10 minute timespan?
>
>Has anyone tried this? I think I remember someone posting a note about this
>some months ago.
>
>Lang
>
--
*--------------------------------------------------*
* ... http://www.trubitt.com *
*--------------------------------------------------*
________________________________________________________________________
________________________________________________________________________
|