Avoiding an overflow in the LTP quantization
authorJean-Marc Valin <jmvalin@jmvalin.ca>
Fri, 22 Jul 2016 19:58:30 +0000 (15:58 -0400)
committerJean-Marc Valin <jmvalin@jmvalin.ca>
Fri, 22 Jul 2016 19:58:30 +0000 (15:58 -0400)
Reduces the XX values to avoid an overflow in silk_VQ_WMat_EC(), even in
floating-point.

silk/tuning_parameters.h

index 1625c3e..d70275f 100644 (file)
@@ -57,7 +57,7 @@ extern "C"
 #define MAX_SUM_LOG_GAIN_DB                             250.0f
 
 /* LTP analysis defines */
-#define LTP_CORR_INV_MAX                                0.015f
+#define LTP_CORR_INV_MAX                                0.03f
 
 /***********************/
 /* High pass filtering */