Hi Jeremy,
Your practical comparisons between MD and DAT are very interesting.
As you know, I did some comparisons on artificial sounds several
months ago:
http://www.avisoft-saslab.com/compression/compression.htm
I agree, that the your samples are nearly identical. However, when I
listen to MDtest2.wav via headphones, I can clearly hear an artifact
(a harsh transient) at t = 6.3 sec, which is not audible at the
corresponding DAT recording at t= 16.4 sec. The spectrograms of both
sections reveal the (to my mind) relatively strong distortion:
DAT:
http://www.avisoft-saslab.com/compression/MDtest2DAT.gif
MD:
http://www.avisoft-saslab.com/compression/MDtest2MD.gif
Especially the additional noise at the beginning of the second
element (6.3 sec) is very evident. Also, some parts of the harmonics
disappeared (but that effect does not seem to be audible).
These findings correspond to the theoretically predicted behavior of
the ATRAC compression algorithm. Each time the signal is too complex
(rapid frequency transients or broad frequency spectra) for the
available bit rate, the algorithm has to discard some information.
This process may cause additional quantization noise and some loss
of relevant parts of the signal.
It is of course very subjective, whether these artifacts are really
audible or not. However, I still have some doubts on the MiniDisk
system. Nevertheless, MiniDisk may be useful for other less complex
natural sounds.
Best regards,
Raimund
--- In Jeremy Minns
<> wrote:
> I 've just started recording with my new Portadisc and am very
pleased with
> it. I had been loath to move from DAT because in spite of all the
postings
> in favour of MD I still had a niggling fear that the quality might
not be
> as good as DAT.
>
> Walter posted a detailed blind comparison of MD v. CD a year or
two ago,
> which showed no audible difference, but since then a few people
have again
> raised doubts as to a possible loss of audio quality due to MD's
ATRAC
> 'compression' . On their web site Cornell write 'recent claims
that Sony's
> ATRAC 3.5 & 4.5 coding versions are considered by some
professionals to be
> indistinguishable from, or even better than a CD remains to be
proven.' I
> thought that as I have both types of recorder I would make a
> small contribution towards resolving the question.
>
> I connected a Sennheiser ME62 in a Telinga reflector to both a
Sony
> TCD-D10PROII (R-DAT) and the Portadisc (MD) and then recorded the
song of a
> Yellow-legged Thrush simultaneously on both recorders. Doug has
posted two
> files on http://www.naturesongs.com/recordists/nrfiles.html ,
MDtest1 and
> MDtest2, which are excerpts from this recording. The only
processing these
> files underwent was that I adjusted the amplitude of each by equal
and
> opposite amounts so that their volume sounded about the same and
converted
> the R-DAT file from 48000 to 41000 samples/sec so that I could
show the
> recordings side by side. Each file consists of the same bit of the
song,
> first as recorded by one recorder and then by the other. The order
was
> decided by spinning a coin.
>
> I'd be interested to know whether you can spot which is which by
ear. If
> you look at the spectograms, however, it becomes fairly obvious,
ATRAC
> having discarded part of the signal, but I think you'll agree that
> difference between the main features of the spectograms is
insignificant. I
> would like to quote to you a 1999 posting to NEOORN by Morton e
Phyllis
> Isler, well known ornithologists working at the Smithsonian, which
puts the
> matter very clearly:
>
> "About a year and a half ago, NEOORN had an excellent back and
forth
> discussion about Minidisks, but a number of you may not have been
enrolled
> at that time, and Charles Duncan's quote from the Budney [curator
of
> Macaulay Library of Natural Sounds] and Grotke [MLNS audio
engineer] paper
> belittling Minidiscs requires a renewed response.
>
> The key question is whether audio engineer Grotke's conclusions
about
> deficiencies in the Minidisk are relevant to its application to
recording
> and using avian vocalizations. Our conclusion, after years of
using
> Minidiscs, is that his conclusions are not relevant. We have also
put
> Nagra, cassette, and DAT recordings directly into CANARY (the
Cornell
> Bioacoustics Workstation) and compared these with the same
recordings put
> onto Minidisks and then into CANARY. We have seen no differences
in the
> displays (e.g. spectrograms) and because we measure vocal
characteristics
> from displays, we can see no difference in vocal measurements that
we
> take. We have never found the slightest evidence of sound
distortion
> caused by the MD."
>
> That sums it up.
>
> The main song in each file is by a male Yellow-legged Thrush. In
MDtest1 a
> Rufous-bellied Thrush calls in the background. In MDtest2 there is
a
> distant call by ( I think) a Tropical Kingbird near the beginning
and a
> Southern House-Wren calls and then sings at the end. The noise in
the
> background is the South Atlantic, not an expressway.
>
> Personally I'm entirely satisfied with MD.
>
> Jeremy
>
>
> [Non-text portions of this message have been removed]
________________________________________________________________________
________________________________________________________________________
|
