Re: [Lame-dev] scalebitcount

[Takehiro T. <tak...@in...>]

From: Robert Hegemann <Rob...@gm...>
Subject: [Lame-dev] scalebitcount
Date: Sun, 29 Feb 2004 11:05:33 +0100

> a quick question:
> will the scale_bitcount and scale_bitcount_lsf functions
> work correct in the case that scalefactors have some
> magic number -1 or -2 ?

No, it's not. These negative numbers need special handling, and they "ARE"
handled properly only in best_scalefac_store().

This is because LAME only calls best_scalefac_store() at the end of
the outer loop and does not call it every time in the inner loop.

Changing this behavior improves encoding quality (theoreticaly), but it
will be very very small, I think.
-- 
Takehiro TOMINAGA // may the source be with you!