This section is obsolete:


[- Please note that this section is very outdated as SecondSpin has been
developed a great deal since I made this and the output quality has been hugely
improved. -]


    Ok. You probably think that doing mpeg stuff with fpu is slower,
    but the mpeg quality is hugely improved if you use floats and its actually
    much faster to use fpu when encoding: _6-7_ times faster in my setup when
    encoding mpeg layer 3.

    The fact is that those free mp3 encoders available which are based to
    the original iso sources aren't that good. I made a test as I had to know
    how much original audio quality suffers when encoding them as mpeg audio:

       I encoded several techno tracks as mpeg audio with LAME and pegase:
         
        - Panasonic  : Kulma         / Tracks #2, #3 (minimalistic techno)
        - Jeff Mills : The Other Day / Track #1      (detroit techno)

       I chose these tracks as they're most likely going to suffer from mpeg
       encoding as they have huge variations of clear noise and dynamics.

       I encoded stereo mpeg layer 3 files with bitrates 128, 160 and 256 plus
       128 kbps j-stereo mpeg layer 2 file with pegase.

       There was no need to use special equipment or programs to analyse the
       audiodata as I think that if you can't hear the difference, WHO CARES!

       When I had encoded the tracks (which was VERY slow, btw) I decoded them
       with mpega (I always use fpu as it has much better quality) to aiff
       files and compiled a audiocd with the encoded tracks as well as the
       originals.

       Then I just started playing with my hifi systems and I couldn't believe
       how much _layer 3_ distorted original Panasonic tracks. They were almost
       unusable! Jeff Mills however was quite good and the low bass and it's
       heavy dynamics were left almost untouched... bitrate didn't seem to
       make much difference (it didn't fix the distortions)..

       The most surprising was that layer 2 actually coped all the tracks
       much better, I'd say almost perfectly. 

       The biggest problem with this test was that as layer 3 lacks j-stereo
       on amiga, so I couldn't test it. Anyway, it shouldn't matter with higher
       bitrates so the test should still be very much referrable.. Other
       problem is that I should test with more complex audio to see if complex
       sound reduces dynamics...

    So here are some hard facts I think you'd like to consider when encoding
    cds to mpeg audio:

        - mpeg layer 3:

            - is much slower than other layers
            - doesn't always have the best audio quality
            - it's the industrial standard but especially the free encoders
              available can deliver _very_ crappy results with certain audiodata
              or when the bitrate isn't high enough.

        - mpeg layer 2 files:

            - files aren't actually much bigger than mp3 and if they are,
              you can always lha them ;)
            - they're easier to decode as well as easier to encode (encoding
              times with those panasonic tracks 2.6:1 mp2, 40:1 mp3!)
            - can actually deliver better results!
            - can't always handle more complex audio correctly (might cut
              bass / treble)

   If you like to encode mpeg audio with your amiga, you should try how
   mpeg layer 2 j-stereo sounds. It's fast and can sometimes give the best
   results.

AG2HTML v2 (c) Lauri Ahonen & Michael Ranner 1996-2000