John Campbell wrote:
>
> Just as we now have Declicking, Debuzzing, Dehissing, Decrackling,
> and so on, I expect that before long we will have DeATRACing.
>
> Like its predecessors, this software won't be able to fully "restore"
> a recording, and it will possibly introduce some by-products of its
> own. But it wll perform well enough in many instances to be regarded
> as a useful or even essential tool. Like other such tools, it will
> interpolate information, based on probabilities.
The first thing that has to be done in creating software is describing
what it's doing precisely. To do that, you have to identify the changes
you want made. To do that you have to find what changes ATRAC made
exactly. Which leads to why this is not likely, it's been very tough to
find anything that ATRAC as a encoding system is doing.
So far all we have moderately certain is that it removes the mic hiss
above about 16 khz. The stuff that's 50dB below the primary sound at
those frequencies. So is the DeATRAC filter just a recording of mic hiss
to add back in? So it would seem, that's going to be easy unless we
demand the same make and model as the original mic. Everybody happy now
that the mic is hissing away properly? Up in the sound frequencies where
your hearing is probably already down 50dB, so you only have to hear the
sound that's effectively, for you, 100 dB down to hear the hiss.
> Why are we creating a situation where such a tool needs to be
> invented? There are several reasons, but chief among these seems to
> be that MiniDisc is a convenient and reliable recording medium. It's
> ironic then that MiniDisc operates on the principle of
> "un-interpolating" based on probabilities. That is, quantisation
> noise will be permitted in certain parts of the spectrum which are
> regarded as "non-critical" in a particular context.
Nice assumption, sorry, no, ATRAC does not do it that way in practice.
Or not so it can be identified. Deciding what ATRAC does using logic and
then trying to find it are like the difference between catching a dead
pig and a stirred up greased one. On a good day if I hold the dowsing
stick just the right way and squint with the right eye, not the left, I
can see clear as day that thing logic predicted. For maybe half a second
before it's gone.
There is no need for such a tool. Start by describing precisely what it
does. You will have to examine lots of recordings as such a tool has to
work across the board. It's going to be interesting waiting for you to
find something that's consistent enough you can identify it and remove
it or add it as the case may be. I have looked for many years now at a
awful lot of stuff recorded with MD.
BTW, every MD recorder has a DeATRACing filter that it applies to the
ATRAC material. It's called the ATRAC decoder section. That does do what
you are asking for. It removes what ATRAC did and restores the original
audio virtually unchanged. But it's a built in part of the ATRAC in the
recorder. Some very talented software/hardware developers have spent may
years perfecting this "filter". And you think someone is going to come
along and do it better? They are welcome to try.
> Unlike earlier eras, where only one medium existed for any
> substantial period of time, and was rapidly made effectively obsolete
> (acoustic phonography, electrical phonography, analogue tape,
> analogue tape plus noise reduction), we now have a choice.
> Uncompressed digital or compressed digital. As I surmise, before
> long we may have the means to psuedo-decompress, and many people
> would be happy with the results - that's assuming they believe that
> process is warranted. Just as, at present, many people are happy
> with their compressed recordings. But given that we do have this
> choice - compressed or uncompressed - I am very curious as to why we
> would not take every opportunity to leave our recordings in the best
> possible state for future generations. If it's not totally essential
> to compromise, then why do it?
Would you instead consider for just a moment some future generation
taking a DAT tape and trying to play it. According to the archive folks
it's risky moments after recording, definitely a problem within a year.
The analog recording methods are still around, why only a few weeks ago
I endured a lecture on how much better a panasonic walkman cassette
recorder was in sound quality than my Portadisc. Several reasons, one of
which was that the cassette did not compress the sound.
> It seems that there are many red herrings introduced into the debate
> over recording media. Yes, the choice of microphone is critical, the
> skill of the operator is an important factor, and so forth. But
> let's be logical and take all these as givens. The only issue is:
> what am I leaving for future listeners, reseachers, whoever? It
> occurs to me that another red herring might be: who is going to care
> about my/your recordings in the future anyway? Is any argument based
> on future action somewhat presumptuous? Yes, of course it is. But
> as we cannot predict how our recordings might be used, why not at
> least ensure that they are not compromised. I don't see how we can
> achieve that using MiniDisc.
They will not in any shape or form be used for critical scientific
analysis of the sound content, that is a red herring too, the biggest
one of all. To have a chance at publishing the research now you have to
have gear with a whole bunch of calibration documentation before even
thinking about analyzing what it produces. You have to have the actual
original recording, not a copy. Often that involves very frequent
recalibration and redocumentation. So don't anybody delude themselves
thinking they are preserving something for analysis. You use pitiful
crude equipment next to those folks, they will sneer at you.
The kinds of uses to which your recording might be suited in science can
be made by a cassette recorder. Undocumented recording, only has to
sound like it should to human ears enough to be identified. Forget
impressing the researchers with your sound quality. At least if you are
going to delude yourself into thinking that you will help science make
sure and buy a good GPS and document where you made the recording. (Even
there, science insists on special, super accurate GPS, with, of course,
it's documentation and calibration trail) I don't want to see, as I have
read on a critical museum specimen tag a location of "around Atlanta".
I'm not as picky as some, if your GPS will only do it within 50 feet
that's ok. Get me that close I can probably find it.
So, predict within the grounds of reality. Ok to listen to, might be
used as documentation of location if it was properly documented and
officially published in a recognized scientific journal. The
documentation and publication is actually the important part, the
recording is just the official voucher, equivalent to the body in the bottle.
And what is this "compromised" garbage. That is not a given. It is pure
speculation with no documentation to back it up.
There is another aspect of this that's completely being ignored. If the
stuff is vanishing and is even now rare to find, it will need lots of
folks out recording. You will get far more recorded then. In order to
get lots of folks to invest and record the gear has to be within their
means. The reality is that you will have a hard time getting them to go
as high as a walkman MD and may have to settle for most of them
recording with cassette Walkman recorders or worse. They will still do
better than one person with super expensive gear will because they will
cover more ground. I've spent the last 5 years recording frogs for
research, and in doing so have had lots of people interested in doing
the same. Very few will even go the few hundred for a homemade parabolic
and a walkman. None, in 5 years has bought anything approaching the
price of a cheap DAT. Except for the very few with the rare research
grant, and most of those bought Portadiscs or MD walkman.
I wonder how many come wandering into this group, read how only a
expensive recorder with a equally expensive mic will do and wander right
back out to take up some cheap hobby, like gold bar collecting. Each of
those folks are recordings not made.
To put it in perspective, the wildlife biologist in charge of the study
I've worked on had a comment when I asked what sort of sound quality he
needed when I first started. He said, more or less, "if you get it
recorded, I can identify it, no matter what the quality, the more you
can give me the better I like it". What he got he loved, even the stuff
recorded with a homemade mic, or a hand held rock band mic next to a
freeway. Which is where I started for him. And some the the stuff he can
pick out of bad recordings is truly amazing. I'm gaining on him and can
do almost as well. You see, what counts to him is getting it down where
these things are, not some pretty recording. He would toss you out there
and ask for more sites if you come in whining about the type of recorder.
I have a area about 1/3 of Georgia. I need enough volunteers to cover
every bit of it from early April to September every night. First one to
find the Pine Barrens Treefrog and properly document it get's their name
permanently enshrined in the scientific journals. That's one of the
tasks I'm working on. BTW, before It's documented I could cook it and
eat it if I wished, or use it for fishbait, it's completely unprotected.
As soon as it's found it will suddenly transform into a endangered
species for Georgia. Even thinking about looking for it will become a
crime if it's like some other states. Fun world we live in. I'm still
debating if I'll report it when first found, or try to get lots of sites first.
> While this applies to recordings in general, I believe it's likely to
> be of utmost relevance to nature recordists. The habitats and
> species we are capturing may not be around forever. If we are going
> to spend considerable time and money doing it, and possibly never
> getting a second chance, let's do it right.
Let's record it on fragile magnetic media, that's the ticket.
The area of documentation of species distribution is the area where
non-scientist nature recordists can do the most good for science. It's a
very time consuming process, very labor intensive, and virtually no
money available to do it. And it's critical to everything else. Getting
the recording in identifiable form regardless of anything is the goal.
It has to be published properly to mean much. Covering lots of ground is
essential. It's rare that sort of work is going to be producing the
sanitized cute recordings we are talking about here, but what it does
produce is worth it's weight in gold. It's going to produce a bunch of
awful mongrel recordings who's only virtue is that the call is
identifiable, and there's a valid record card. Yes, once in a while you
will get a good site and a wonderful recording, before dashing for the
next site. It's parallel, BTW is in my collection of photo documentation
for the survey. Can you visualize some sort of half ground up animal on
the road, most commonly a snake, some are still partially alive, that's
my average survey photo. You see the lucky ones on my website.
For those that like more exotic locations, Doug once asked me to
identify the frogcalls he was getting in Costa Rica. I have not yet
found a guide to the calls and there appears not to be one. I have a
list of the species in Costa Rica, there are 125 species known there.
For those who want a calling, all you have to do is record the frog,
catch the frog, identify the frog, and document/publish for each of
these. That would be a definite contribution. BTW, it does not have to
be Costa Rica, there's plenty of countries for all.
> I'm gonna get me a damn good red herring recording before them
> critters mutate into permanantly pickled herrings.
You'll need a hydrophone. (and don't forget the GPS readings) Me, I'm
too busy documenting the expansion of the known range of frog species in
Georgia. To keep other scientists from doing it by pickling them to put
in jars in a museum.
Walt
________________________________________________________________________
________________________________________________________________________
|