eden-emu/opus
15
0
Fork 0
mirror of https://github.com/xiph/opus.git synced 2025-05-30 07:07:42 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fixes an unstable energy issue for low-bitrate hybrid

Browse source 
The transient detector would trigger on low-pitch vowels, but we didn't have
enough bits to properly code the high bands as a transient, resulting in
partial collapse and unstable energy.
This commit is contained in:
Jean-Marc Valin 2016-10-27 14:14:28 -04:00
parent 76fa939bde
commit 2af92cd99f
No known key found for this signature in database
GPG key ID: 5E5DD9A36F9189C8
1 changed files with 23 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

27
celt/celt_encoder.c
View file

@ -226,7 +226,7 @@ void opus_custom_encoder_destroy(CELTEncoder *st)
static int transient_analysis(const opus_val32 * OPUS_RESTRICT in, int len, int C, static int transient_analysis(const opus_val32 * OPUS_RESTRICT in, int len, int C,
opus_val16 *tf_estimate, int *tf_chan) opus_val16 *tf_estimate, int *tf_chan, int low_rate)
{ {
int i; int i;
VARDECL(opus_val16, tmp); VARDECL(opus_val16, tmp);
@ -236,6 +236,12 @@ static int transient_analysis(const opus_val32 * OPUS_RESTRICT in, int len, int
int c; int c;
opus_val16 tf_max; opus_val16 tf_max;
int len2; int len2;
/* Forward masking: 6.7 dB/ms. */
#ifdef FIXED_POINT
int forward_shift = 4;
#else
opus_val16 forward_decay = QCONST16(.0625f,15);
#endif
/* Table of 6*64/x, trained on real data to minimize the average error */ /* Table of 6*64/x, trained on real data to minimize the average error */
static const unsigned char inv_table[128] = { static const unsigned char inv_table[128] = {
255,255,156,110, 86, 70, 59, 51, 45, 40, 37, 33, 31, 28, 26, 25, 255,255,156,110, 86, 70, 59, 51, 45, 40, 37, 33, 31, 28, 26, 25,
@ -250,6 +256,18 @@ static int transient_analysis(const opus_val32 * OPUS_RESTRICT in, int len, int
SAVE_STACK; SAVE_STACK;
ALLOC(tmp, len, opus_val16); ALLOC(tmp, len, opus_val16);
/* For lower bitrates, let's be more conservative and have a forward masking
decay of 3.3 dB/ms. This avoids having to code transients at very low
bitrate (mostly for hybrid), which can result in unstable energy and/or
partial collapse. */
if (low_rate)
{
#ifdef FIXED_POINT
forward_shift = 5;
#else
forward_decay = QCONST16(.03125f,15);
#endif
}
len2=len/2; len2=len/2;
for (c=0;c<C;c++) for (c=0;c<C;c++)
{ {
@ -302,9 +320,9 @@ static int transient_analysis(const opus_val32 * OPUS_RESTRICT in, int len, int
mean += x2; mean += x2;
#ifdef FIXED_POINT #ifdef FIXED_POINT
/* FIXME: Use PSHR16() instead */ /* FIXME: Use PSHR16() instead */
tmp[i] = mem0 + PSHR32(x2-mem0,4); tmp[i] = mem0 + PSHR32(x2-mem0,forward_shift);
#else #else
tmp[i] = mem0 + MULT16_16_P15(QCONST16(.0625f,15),x2-mem0); tmp[i] = mem0 + MULT16_16_P15(forward_decay,x2-mem0);
#endif #endif
mem0 = tmp[i]; mem0 = tmp[i];
} }
@ -314,6 +332,7 @@ static int transient_analysis(const opus_val32 * OPUS_RESTRICT in, int len, int
/* Backward pass to compute the pre-echo threshold */ /* Backward pass to compute the pre-echo threshold */
for (i=len2-1;i>=0;i--) for (i=len2-1;i>=0;i--)
{ {
/* Backward masking: 13.9 dB/ms. */
#ifdef FIXED_POINT #ifdef FIXED_POINT
/* FIXME: Use PSHR16() instead */ /* FIXME: Use PSHR16() instead */
tmp[i] = mem0 + PSHR32(tmp[i]-mem0,3); tmp[i] = mem0 + PSHR32(tmp[i]-mem0,3);
@ -1586,7 +1605,7 @@ int celt_encode_with_ec(CELTEncoder * OPUS_RESTRICT st, const opus_val16 * pcm,
if (st->complexity >= 1 && !st->lfe) if (st->complexity >= 1 && !st->lfe)
{ {
isTransient = transient_analysis(in, N+overlap, CC, isTransient = transient_analysis(in, N+overlap, CC,
&tf_estimate, &tf_chan); &tf_estimate, &tf_chan, effectiveBytes<15);
} }
if (LM>0 && ec_tell(enc)+3<=total_bits) if (LM>0 && ec_tell(enc)+3<=total_bits)
{ {