[78-L] Reversing audio

[Michael Biel]

Confusion has arisen here because the original context of the situation 
was not included in the copies in the thread, and therefore has not 
become a part of the experiments being undertaken. The question was not 
about simply reversing audio which had originated in the digital domain. 
The situation is about analog audio which was initially reversed while 
in the analog domain. It was about digitally reversing an analog 
recording which had been reversed in the analog domain and digitalized 
while being played backwards. When you then return it to analog (which 
has to happen when playing it) and compare it with the original analog 
recording, according to the presenter they don't match. The square waves 
mentioned in the presentation were not generated in the digital domain 
like David Burnham's, they were analog tape recordings of a square wave 
that were reversed in the analog domain and then re-reversed in the 
digital domain and compared with the original analog recording.

This discussion started about playing an LP in reverse when 
digitalizing. The ARSC presentation concerned digitalizing 
bi-directionally recorded tapes in one pass, with some tracks being 
played forwards while other tracks were being played backwards and being 
restored to the proper direction in the digital domain.

So now repeat your experiments using your A-D and D-A converters in the 
chain.


Doug Pomeroy wrote:

 > I assume Mike is referring to Mark Hood's talk? I missed it (will 
watch when it is
 > finally posted on the ARSC web site)

I am about to sit down to my videotape of the talk to try to figure out 
the confusion myself)

 > but I spoke to Mark about his talk (and his
 > examination of waveforms produced by playing analog tapes in reverse) 
but he
 > didn't mention anything about digitally reversing audio,

I don't know what he told you, but if you look at his abstract in the 
convention guide, he clearly includes digital reversal of reversed 
analog tape recordings. I've separated that part out into a second 
paragraph. Note his final sentence which was a hint to what he was going 
to present, unfortunately without giving away his findings in advance.


 >>> POLARITY AND PHASE RESPONSE IN REVERSE PLAYBACK OF ANALOG MAGNETIC TAPE
Mark Hood, Indiana University Archives of Traditional Music, 
Bloomington, Indiana

 >>> Reproducing the content of analog tapes “backwards” was long a part 
of standard practices
in dubbing tapes for broadcast distribution and the mass-production of 
certain configurations
of pre-recorded consumer tapes. In addition to increasing efficiency by 
allowing
duplicators to dub multiple audio program streams simultaneously (e.g., 
the “forward”
A side and the “backward” B side of a cassette), this practice was also 
shown to
prevent the serial accumulation of phase distortion products inherent in 
the magnetic tape
recording process itself.

 >>> In the current era, audio archivists may be able to achieve a
major increase in productivity when digitizing tapes that contain 
multiple tracks of audio
content that were recorded bi-directionally by playing back these tapes 
on reproducers
with head configurations that retrieve all of the content 
simultaneously. Preservation
files created by this method that contain “backward” audio can then be 
processed in the
digital domain by a quick and simple re-ordering of the samples to 
create a file that plays
in a normal, forward manner. As part of the Sound Directions project, 
the Archives of
Traditional Music at Indiana University is conducting research into 
increasing efficiency
and throughput by employing various methods of parallel transfer 
(simultaneous digitization
of multiple audio streams) and workflow automation. Our investigation 
into the use
of reverse tape playback has uncovered some interesting results that may 
have bearing on
the use of this technique in audio archival practice.<<<


 > and I cannot imagine why reversing the order of digital samples would
 > have any deleterious effects. Bits is bits, no? Doug Pomeroy 
audiofixer at verizon.net

I assume your last question was meant to be humorous, but actually in audio it has been determined that bits is not necessarily bits because the resultant restoration to analog can be disturbed by jitter and other clocking errors.  They are further affected by error correction and error concealment.  All of our pre-digital recordings have to pass through an A-D stage and since things can't remain in the digital domain forever if we want to hear them, a D-A stage as well.  

Mike Biel  mbiel at mbiel.com