[opus] [PATCH] Preloaded to stack array LPC is used.

[Cyril Lashkevich]

It is passed to silk_LPC_analysis_filter instead of A_Q12. According
to comment preloading to stack is done for performance gain, so
preloaded LPC used everywere.
---
 silk/decode_core.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/silk/decode_core.c b/silk/decode_core.c
index 0365ffd..e1e3206 100644
--- a/silk/decode_core.c
+++ b/silk/decode_core.c
@@ -147,7 +147,7 @@ void silk_decode_core(
                 }
 
                 silk_LPC_analysis_filter( &sLTP[ start_idx ], &psDec->outBuf[ start_idx + k * psDec->subfr_length ],
-                    A_Q12, psDec->ltp_mem_length - start_idx, psDec->LPC_order );
+                    A_Q12_tmp, psDec->ltp_mem_length - start_idx, psDec->LPC_order );
 
                 /* After rewhitening the LTP state is unscaled */
                 if( k == 0 ) {
-- 
1.7.10.2 (Apple Git-33)