Thanks for the summary, Mark.
I forgot to mention the 330/332 bug, which I had independently located and
saw discussed here earlier. I looked at the ISO code for the Huffman
quadruples, and it looks OK. There is very little scfsi related code, and
it looks OK, though I don't think I've ever seen it used by an encoder, so I
can't say that it's been thoroughly tested.
As for switching rates (or CRC protection or what have you in the header),
our decoder is used for playing back multicast data - it has to be able to
handle on-the-fly changes, as there aren't any other kind for us. Decoders
that can't handle such changes, BTW, are not up to spec. It really isn't
that hard to handle them - you just can't make assumptions about the whole
stream from the first frame, which is a bad idea anyway, as someone was
noting today in the case of CRC protection, and is doubly stupid if you're
streaming, as an error in the first frame will completely hose you. Of
course making such assumptions can allow some (usu. minor) speedups, on the
other hand.
> There is a Word document floating around, published by
> Fraunhofer, called "MPEG Layer3 Bitstream syntax and decoding"
> It is much more usefull to decoder writers than the ISO docs.
Anybody have a copy of this? I'd love to see it.
Thanks,
Alex
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
