[Les Cargill[_4_]]

Peter Larsen wrote:
alex wrote:

[mp3 encode-decode]

Other than this the encoding process tend to make the waveform
smoother and less compex, which, sometimes can appear as a sound
"improvement"...

I have a box with 100 big-band numbers. It is better when mp3'ed because
that hides some of the noise-reduction artifacts.


Weird. You mean companding artifacts? Not something I would
have expected. I have a few mixes from a Tascam 488MkII
8 track cassette with dbx, and the .mp3s fmro that
sound a lot like the .wav files.


The worst issue with mp3 encode-decode is poor stereo encoding choices, I
haven't seen other encoders than Audition's allow easy access to those, but
nor have I looked very much. Basically wise choices allow you to preserve
stereo, bad choices makes you loose it.

I find it useful to downsample to 32 kHz sample rate prior to encoding, it
reduces the splattyness that comes from what appears to be the replacement
of treble with white noise and allows max quality variable bandwindth
encoding to do its best at lowest quality cost. The result is reasonable
"compact cassette type quality".


I get bigger error-difference-files from samplerate conversion ( using
CoolEdit 96 ) than from MP3 conversion.

Kind regards

Peter Larsen



--
Les Cargill