>
> Hello,
>
> I believe Robert said that mp3 frames overlap 50%, then would it be
> sufficient to init some values using _only_ the previous frame in
> order to play the next ok?
>
> So: I want good playback, starting with frame N, would it suffice to
> load up frame N-1, and then start playing @ frame N? (or do I need
> more frames (N-2, ...))
>
> (I'm wondering to which frame to refer in the albumheader)
>
> thanks
>
> --
> Best regards,
> Roel mailto:[EMAIL PROTECTED]
>
Do you want absolutely perfect reconstruction of frame N?
To fully decode frame N, you need to:
uncompress frames frame N-1,N and N+1
sum data (with a 50% overlap) to get 1152 samples from frame N
But to uncompress frame N-1, 99.9% of the time, the data for this
frame will be partially stored in frame N-2.
So say you start with frame N-1:
1. Most of the time, Frame N-1 will not be uncompressed since it needs data
from frame N-2.
2. Frame N will be properly uncompressed. But when you sum
it with the output of frame N-1 (because of the 50% overlap)
some of the data will be corrupted since frame N-1 was not
properly decoded. (the first
288 samples will be corrupted depending on what the decoder
uses from frame N-1.)
3. Frame N+1 will be properly uncompressed and fully decoded.
And what about the other .1% of the time? In some cases,
(if bitreservoir=0) frame N-1 will be uncompressed, and then
frame N-1 will be output (with some corrupted data) and
frame N will be okay.
But you can also have the situation where the bitreservoir
is quite large - I think these are the worst case situations:
MPEG1: data for frame N can be in frame N-8
MPEG2: data for frame N can be in frame N-256!
But you can only get these extreme numbers by encoding
pure digital silence at the lowest possible bitrate.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
