From d884af2b9992f12e98a3e8548aff76b232b5bfb3 Mon Sep 17 00:00:00 2001 From: Mohamed Tarek Date: Mon, 5 Jul 2010 22:33:37 +0000 Subject: Partial conversion of the wmapro decoder to fixed point arithmetic. Currently inverse quantization & rescaling, imdct and windowing are all in fixed point. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@27302 a1c6a512-1295-4272-9138-f99709370657 --- apps/codecs/libwmapro/README.rockbox | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) (limited to 'apps/codecs/libwmapro/README.rockbox') diff --git a/apps/codecs/libwmapro/README.rockbox b/apps/codecs/libwmapro/README.rockbox index 09c54b48c6..9bd737a432 100644 --- a/apps/codecs/libwmapro/README.rockbox +++ b/apps/codecs/libwmapro/README.rockbox @@ -14,8 +14,18 @@ IMPORT DETAILS Based on ffmpeg svn r22886 dated 15 April 2010. -Currently, the files contain minimal changes from their original state in order -to be able to compile cleanly. +The code is slowly being modified to convert it from floating point maths to +fixed point. + +As of 6 July 2010, the following steps are all working properly in fixed point: +- Inverse quantization and rescaling +- IMDCT +- Windowing + +Results of comparing output wav files from the partially fixed point decoder to +the output wav files of ffmpeg decoder are : +- average relative error = 0.016% +- maximum relative error = 0.3% COMPILING -- cgit v1.2.3