[FFmpeg-soc] [soc]: r4856 - amr/amrnbdec.c

[cmcq]

Author: cmcq
Date: Fri Jul 31 05:08:18 2009
New Revision: 4856

Log:
Use multiply instead of divide, for speed

Modified:
   amr/amrnbdec.c

Modified: amr/amrnbdec.c
==============================================================================
--- amr/amrnbdec.c	Fri Jul 31 05:05:57 2009	(r4855)
+++ amr/amrnbdec.c	Fri Jul 31 05:08:18 2009	(r4856)
@@ -916,7 +916,7 @@ static int synthesis(AMRContext *p, floa
     // factor of 4
     if (overflow)
         for (i = 0; i < AMR_SUBFRAME_SIZE; i++)
-            p->pitch_vector[i] /= 4.0;
+            p->pitch_vector[i] *= 0.25;
 
     // construct the excitation vector
     for (i = 0; i < AMR_SUBFRAME_SIZE; i++)