Re: MP3 conversions disc for IIGS?

[mjmahon@aol.com (Michael J. Mahon)]

Supertimer wrote:

>The file won't be an MP3 anymore.  Oversampler replaces the
>MP3 compression with interpolation.  The file is first converted
>to a standard wave file on the PC which is then reduced in size
>by downsampling to 26360Hz.  On the IIGS side, Oversampler
>then "oversamples" the file at either 2x or 4x to gain back
>quality while maintaining a reduced file size.

Interpolating to upsample adds exactly zero information to
the output, so what exactly do you mean by "gain back
quality"?

When downsampling, you must digital filter out everything
above the Nyquist frequency, or half the sample frequency,
to avoid power at those frequencies from aliasing back down
below the Nyquist frequency and causing distortion.  Re-up-
sampling does not undo this folding, and cannot restore any
lost information.

The only thing which upsampling can accomplish is to raise
the frequency of the sampling noise, which, at 26KHz is
already inaudible to 99.7% of the population--or 100% of
the population that has attended a rock concert.  ;-)  So I
can't see this as a viable reason to upsample in playback.

>Another option would be to take the 44.1kHz wave file itself
>and save that with the right attributes for the IIGS and use
>Oversampler to play it at 1x sample rate.  However, because
>the file no longer has MP3 compression, it will be very large
>this way and as mentioned in my article, the IIGS power
>supply has interference.  In my experimentation, I have found
>that in order to make the 44.1kHz file sound right, you would
>have to upsample the file to 52720Hz (2x 26360Hz).  This
>will give you the best sound quality at the expense of even
>more disk space.

What do you mean by "sound right"?  Is there sufficient noise
from the switching power supply (whose frequency is almost
certainly not well controlled) that the beat with the sampling
frequency is audible when the sampling frequency is 22.05KHz
or 44.1KHz.  (The lowest beat frequency would be 4310Hz and
8620Hz, respectively, but should be of relatively low amplitude
if any care is taken in power supply filtering to the audio circuits.)
Is ultrasonic power supply noise an issue on all IIgs's, or just
some?

If power filtering _is_ the issue, then the addition of a 47mf capacitor
from +5 to ground near the audio circuits should markedly decrease
the problem in a way that would improve _all_ sampled sound on
the IIgs in question.

If you're a purist, then the fractional upsampling must be done
very carefully (more carefully than linear interpolation) to avoid
introducing some low-level distortion, but I suspect that most
people playing back 8-bit sampled music on their IIgs's are not
"purists".




-michael

 Email:  mjmahon@aol.com
 Home page:  http://members.aol.com/MJMahon/