RE: More ranting from me!

[Nigel Orr <Nigel.Orr@xxxxxxx>]

At 12:30 20/06/00 +0100, you wrote:
>don't know which track is which. I think I might have learned something
>about mp3 here - I take it that the tracks are not reconstructed during
>playback as they are with Meridian Lossless packing, so I can play them
back
>through any HiFi?

Dunno about the Meridian system, presumably it's just lossless data
compression?

Basically, when you play an MP3 back, the sound can come out the speakers
or go directly to a .wav file, which will be exactly the same size as the
original .wav file, but will have a slightly different waveform.

MP3 encoding uses various techniques to chuck away 'imperceptible' sound,
and store the result in a standard file format (eg sounds below the
threshold of audibility, sounds perceptually masked by other
nearby-in-frequency sounds etc etc).  Once it's decoded, you should have
the original sound, minus any theoretically imperceptible components, plus
any processing artefacts which might have sneaked in.  The higher the bit
rate, the less has to be chucked away to fit in the space provided.

That's why testing MP3 encoders with white noise or test tones just doesn't
work, test tones will probably be recreated perfectly, white noise will
probably be butchered.  At the end of my pile of CDs are a couple of
technical test ones- I'll encode them too just to hear what the various
signals sound like after MP3'ing...

And it's also why another technique, of listening to the difference between
the original source and the encoded signal, is meaningless, because what
the bits that are thrown away sound like is completely irrelevant, the only
important thing is what their absence sounds like from the original
material- confusing maybe...

>Not having played with it yet though I would imagine that
>if you burn to an Audio CD the file sizes are the same as the original
>uncompressed sample?

Yes.  But you'd be surprised how many people think it shouldn't be...

>Lastly recommendations for a good encoder that I can download and use
for
>this experiment?

Dunno for Windows (no, that's not the name of an encoder!)... but you
probably have to pay for the best ones...

Have a look at the http://www.mp3-tech.org site that
someone pointed out a
few days ago- there are various test results there at various bit rates
(and with assorted encoder systems, not just MP3), and details of the
tests, so you can recreate them.  If you want to listen to the MP3s on
cheap speakers with a soundcard, then do the tests like that, if you're
going to listen to them on a decent hi-fi, do the tests on that.  Some MP3
artefacts might sound nicer on one system than the other...

Nigel


------------------------------------------------------------------------
BeMANY, where eGroups members SAVE on long distance.
http://click.egroups.com/1/4121/7/_/2065/_/961502927/
------------------------------------------------------------------------