diff options
| author | Linfeng Zhang <linfengz@google.com> | 2016-07-13 17:25:49 -0700 |
|---|---|---|
| committer | Jean-Marc Valin <jmvalin@jmvalin.ca> | 2017-02-14 23:57:15 -0500 |
| commit | 95d4c9f960c9469961781c923ccfdb8c3eba0562 (patch) | |
| tree | 293fb4a4794c6dc74d7ed7381a543e88f741a8c9 /silk/PLC.c | |
| parent | 68688651a4c3ed1fc4345c1bfb3932658e51f0b4 (diff) | |
| download | opus-95d4c9f960c9469961781c923ccfdb8c3eba0562.tar.gz | |
Optimize silk_LPC_inverse_pred_gain() for ARM NEON
The optimization is bit exact with C function.
Change-Id: Ib3bdc26a5a4ebe02e7f24be85104e8e9a2a9a738
Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca>
Diffstat (limited to 'silk/PLC.c')
| -rw-r--r-- | silk/PLC.c | 2 |
1 files changed, 1 insertions, 1 deletions
@@ -275,7 +275,7 @@ static OPUS_INLINE void silk_PLC_conceal( /* Reduce random noise for unvoiced frames with high LPC gain */ opus_int32 invGain_Q30, down_scale_Q30; - invGain_Q30 = silk_LPC_inverse_pred_gain( psPLC->prevLPC_Q12, psDec->LPC_order ); + invGain_Q30 = silk_LPC_inverse_pred_gain( psPLC->prevLPC_Q12, psDec->LPC_order, arch ); down_scale_Q30 = silk_min_32( silk_RSHIFT( (opus_int32)1 << 30, LOG2_INV_LPC_GAIN_HIGH_THRES ), invGain_Q30 ); down_scale_Q30 = silk_max_32( silk_RSHIFT( (opus_int32)1 << 30, LOG2_INV_LPC_GAIN_LOW_THRES ), down_scale_Q30 ); |