> Date: Tue, 11 Jul 2000 14:37:35 -0700
> From: "Chris Haynes" <[EMAIL PROTECTED]>
>
> We know that LAME now has roughly the same quality of MP3Enc 3.1. But,
> as far as I am concerned, full huffman search hasn't been implemented
> on LAME yet. I've noticed LAME 3.8x produces better quality than 3.70,
> and I presume the main reason is the more efficient huffman coding of
> Takehiro. MP3Enc does full huffman search on qualilty 9. So, should MP3Enc
> with -qual 9 switch sound a bit better than LAME -h? Please let me know.
>
Right now LAME does do a full huffman search after the outer_loop
iteration, and the bits that this saves are available for the next
granule. But it is also possible to do a full huffman search for
every iteration - and I assume this is what you mean?
This work should be done, but: it will be very slow for a small gain.
Also, it will only help CBR. ABR and VBR do not need this, since they
dont have rigid bitrate constraints. For example, with ABR we could
just increase the target bitrate by 5% before the outer_loop() call,
assuming that after outer_loop() is done, best_huffman_divide is going
to save us 5%.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
