naturerecordists
[Top] [All Lists]

Re: 24/96

Subject: Re: 24/96
From: Walter Knapp <>
Date: Wed, 23 Apr 2003 13:19:49 -0400
Raimund Specht wrote:

> I know some of these discussions on MiniDisk from the journal of the 
> Wildlife Sound Recording Society (e.g. Vol 8, No 5, spring 1999). 
> Also, I do not want to argue against the use of MiniDisk for nature 
> sound recording. Most of the nature sound recordists record for 
> their own listening pleasure and there is no doubt, that ATRAC is 
> appropriate for this purpose.

I'm tired of this put down of nature recordists. The most expert nature 
recordists are not necessarily scientists. Scientists are often rank 
amateurs in recording.

Recording for science makes allowances for all kinds of bad things in 
recordings. Or simply ignores them. Recording for listening pleasure is 
by far the most challenging.

> It's also true, that the effects caused by ATRAC are often not 
> relevant. Other effects might be more critical for spectrographic 
> analysis (e.g. reverberation and background noise). However, if the 
> recording conditions are good, the drawbacks of the compression may 
> become visible in complex sounds. The temporal masking effects occur 
> on a time scale of a few milliseconds only. If you look at a 
> spectrogram with a large FFT length (say 1024 or more), the temporal 
> resolution of this spectrogram would not be appropriate to see these 
> effects. Imagine you had a soundfile with a sample rate of 22.05 
> kHz, a FFT size of 1024 and a Hamming window. This configuration 
> will provide a temporal resolution of 36 milliseconds only (analysis 
> bandwidth = 28 Hz). Obviously, we would be unable to see the effects 
> we are looking for. Instead we had to use a much larger analysis 
> bandwidth. A FFT length of 128 (or zero-padding on larger FFT sizes) 
> would resolve temporal details down to 4 milliseconds (analysis 
> bandwidth = 220 Hz). That would be more appropriate and one should 
> see some distortion in rapidly modulated signals. However, if you 
> look at smooth whistles, there will be probably no major 
> degeneration.

I've been waiting for 7 years for someone to actually show these effects 
that are so far just imaginary, nothing more than a hypothesis. I have 
done considerable looking on my own as well, I'm a scientist. It's time 
someone put up or shut up. All we have had is Cornell trying to pass off 
digital clipping as the effect of ATRAC. Still trying to pass it off as 
the effect of ATRAC today on their website. That's how frantic some are 
to discredit ATRAC. I don't understand this sillyness, it's definitely 
not scientific. A scientist knows that his hypothesis must be tested or 
it's nothing. A scientist provides real data and shows how that's relevant.

So, do it, show us, and demonstrate how it makes a difference to the 
animals. I've done every resolution of sonogram available to me without 
finding what you talk of. That's lengths from 128 to 4096. I've looked 
at great lengths of ATRAC recorded sound on a wave by wave basis as well 
in the course of editing recordings. The call to call variation of one 
individual is very obvious on sonograms. If what you are talking about 
is there in even smaller form it's going to get lost in that call to 
call variation.

I for one do not care particularly if there is some variation in part of 
one wave of sound because it has not been shown to be relevant. At 1kHz 
you are talking about a hypothesis of changing a group of just a few 
cycles with a 4 millisecond timeframe. It's trivia, definitely not 
important to the animals. If it was all that important then the fact you 
defined each wave with a small number of samples is also important. And 
that tosses all digital recording ultimately. And if the animals were 
that insistent on such perfection of call formation, there would be no 
breeding, as their calls are not that perfect. Even if they could 
produce absolutely perfect calls one after another the atmospheric 
variations would change each call in different ways. Ways that cannot be 
calculated out of the data.

I was reminded of that last night trying to triangulate a screech owl 
call. At such a distance that the thermal variations in the atmosphere 
were breaking it up.

You might want to go read Tinbergen's experiments with herring gulls. 
Just how little of the details of the adult bird's head were necessary 
for the chick to recognize it. I'm sure the same thing applies to sound. 
Animal Ethology was a interesting course when I took it. You look into 
this sort of stuff there. That is the bottom line in Bioacoustics, what 
matters to the animals.

I have recorded using some of the finest mics in near ideal conditions 
using ATRAC. I'm sorry to report, there is still no masking, no obvious 
errors, not even minor ones of the nature claimed that can be attributed 
to ATRAC. In fact such conditions are simpler than less ideal conditions 
with lots of extra noises, so easier to manipulate. The less ideal 
conditions are more challenging, containing much more variation in 
signal. And there is no masking there either. It would be nice to have a 
switch on a recorder where we could mask out faint background sounds 
that are unwanted. But ATRAC won't do that.

Walt





________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU