cpet/openbsd-ports
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Back port some more AAC encoder fixes.

Browse Source 
from Brad (maintainer)
This commit is contained in:
ajacoutot 2015-09-15 12:49:42 +00:00
parent 2be6e94ebb
commit f6549ec49f
3 changed files with 225 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
graphics/ffmpeg/Makefile
View File

@ -1,11 +1,11 @@
# $OpenBSD: Makefile,v 1.124 2015/09/12 09:44:54 ajacoutot Exp $
# $OpenBSD: Makefile,v 1.125 2015/09/15 12:49:42 ajacoutot Exp $
COMMENT= audio/video converter and streamer
V= 20150909
DISTNAME= ffmpeg-git-${V}
PKGNAME= ffmpeg-${V}
REVISION= 0
REVISION= 1
CATEGORIES= graphics multimedia
MASTER_SITES= http://comstyle.com/source/
EXTRACT_SUFX= .tar.xz

191
graphics/ffmpeg/patches/patch-libavcodec_aacenc_tns_c Normal file
View File

@ -0,0 +1,191 @@
$OpenBSD: patch-libavcodec_aacenc_tns_c,v 1.1 2015/09/15 12:49:42 ajacoutot Exp $
aacenc_tns: redo coefficient quantization and decision making
aacenc_tns: encode coefficients directly and reenable compression
--- libavcodec/aacenc_tns.c.orig Sat Sep 12 15:01:30 2015
+++ libavcodec/aacenc_tns.c Sat Sep 12 15:02:10 2015
@@ -31,17 +31,34 @@
#include "aacenc_utils.h"
#include "aacenc_quantization.h"
+/*
+ * Shifts the values as well if compression is possible.
+ */
+static inline int compress_coeffs(int *coef, int order, int c_bits)
+{
+ int i, res = 0;
+ const int low_idx = c_bits ? 4 : 2;
+ const int shift_val = c_bits ? 8 : 4;
+ const int high_idx = c_bits ? 11 : 5;
+ for (i = 0; i < order; i++)
+ if (coef[i] < low_idx && coef[i] > high_idx)
+ res++;
+ if (res == order)
+ for (i = 0; i < order; i++)
+ coef[i] -= (coef[i] > high_idx) ? shift_val : 0;
+ return res == order;
+}
+
/**
* Encode TNS data.
* Coefficient compression saves a single bit per coefficient.
*/
void ff_aac_encode_tns_info(AACEncContext *s, SingleChannelElement *sce)
{
- uint8_t u_coef;
- const uint8_t coef_res = TNS_Q_BITS == 4;
int i, w, filt, coef_len, coef_compress = 0;
const int is8 = sce->ics.window_sequence[0] == EIGHT_SHORT_SEQUENCE;
TemporalNoiseShaping *tns = &sce->tns;
+ const int c_bits = is8 ? TNS_Q_BITS_SHORT == 4 : TNS_Q_BITS == 4;
if (!sce->tns.present)
return;
@@ -49,38 +66,24 @@ void ff_aac_encode_tns_info(AACEncContext *s, SingleCh
for (i = 0; i < sce->ics.num_windows; i++) {
put_bits(&s->pb, 2 - is8, sce->tns.n_filt[i]);
if (tns->n_filt[i]) {
- put_bits(&s->pb, 1, coef_res);
+ put_bits(&s->pb, 1, c_bits);
for (filt = 0; filt < tns->n_filt[i]; filt++) {
put_bits(&s->pb, 6 - 2 * is8, tns->length[i][filt]);
put_bits(&s->pb, 5 - 2 * is8, tns->order[i][filt]);
if (tns->order[i][filt]) {
+ coef_compress = compress_coeffs(tns->coef_idx[i][filt],
+ tns->order[i][filt], c_bits);
put_bits(&s->pb, 1, !!tns->direction[i][filt]);
put_bits(&s->pb, 1, !!coef_compress);
- coef_len = coef_res + 3 - coef_compress;
- for (w = 0; w < tns->order[i][filt]; w++) {
- u_coef = (tns->coef_idx[i][filt][w])&(~(~0<<coef_len));
- put_bits(&s->pb, coef_len, u_coef);
- }
+ coef_len = c_bits + 3 - coef_compress;
+ for (w = 0; w < tns->order[i][filt]; w++)
+ put_bits(&s->pb, coef_len, tns->coef_idx[i][filt][w]);
}
}
}
}
}
-static inline void quantize_coefs(double *coef, int *idx, float *lpc, int order)
-{
- int i;
- uint8_t u_coef;
- const float *quant_arr = tns_tmp2_map[TNS_Q_BITS == 4];
- const double iqfac_p = ((1 << (TNS_Q_BITS-1)) - 0.5)/(M_PI/2.0);
- const double iqfac_m = ((1 << (TNS_Q_BITS-1)) + 0.5)/(M_PI/2.0);
- for (i = 0; i < order; i++) {
- idx[i] = ceilf(asin(coef[i])*((coef[i] >= 0) ? iqfac_p : iqfac_m));
- u_coef = (idx[i])&(~(~0<<TNS_Q_BITS));
- lpc[i] = quant_arr[u_coef];
- }
-}
-
/* Apply TNS filter */
void ff_aac_apply_tns(AACEncContext *s, SingleChannelElement *sce)
{
@@ -122,22 +125,41 @@ void ff_aac_apply_tns(AACEncContext *s, SingleChannelE
}
}
+/*
+ * c_bits - 1 if 4 bit coefficients, 0 if 3 bit coefficients
+ */
+static inline void quantize_coefs(double *coef, int *idx, float *lpc, int order,
+ int c_bits)
+{
+ int i;
+ const float *quant_arr = tns_tmp2_map[c_bits];
+ for (i = 0; i < order; i++) {
+ idx[i] = quant_array_idx((float)coef[i], quant_arr, c_bits ? 16 : 8);
+ lpc[i] = quant_arr[idx[i]];
+ }
+}
+
+/*
+ * 3 bits per coefficient with 8 short windows
+ */
void ff_aac_search_for_tns(AACEncContext *s, SingleChannelElement *sce)
{
TemporalNoiseShaping *tns = &sce->tns;
int w, w2, g, count = 0;
const int mmm = FFMIN(sce->ics.tns_max_bands, sce->ics.max_sfb);
const int is8 = sce->ics.window_sequence[0] == EIGHT_SHORT_SEQUENCE;
- const int order = is8 ? 7 : s->profile == FF_PROFILE_AAC_LOW ? 12 : TNS_MAX_ORDER;
+ const int c_bits = is8 ? TNS_Q_BITS_SHORT == 4 : TNS_Q_BITS == 4;
int sfb_start = av_clip(tns_min_sfb[is8][s->samplerate_index], 0, mmm);
int sfb_end = av_clip(sce->ics.num_swb, 0, mmm);
for (w = 0; w < sce->ics.num_windows; w++) {
- float e_ratio = 0.0f, threshold = 0.0f, spread = 0.0f, en[2] = {0.0, 0.0f};
- double gain = 0.0f, coefs[MAX_LPC_ORDER] = {0};
+ int use_tns;
+ int order = is8 ? 5 : s->profile == FF_PROFILE_AAC_LOW ? 12 : TNS_MAX_ORDER;
int coef_start = w*sce->ics.num_swb + sce->ics.swb_offset[sfb_start];
int coef_len = sce->ics.swb_offset[sfb_end] - sce->ics.swb_offset[sfb_start];
+ float e_ratio = 0.0f, threshold = 0.0f, spread = 0.0f, en[2] = {0.0, 0.0f};
+ double gain = 0.0f, coefs[MAX_LPC_ORDER] = {0};
for (g = 0; g < sce->ics.num_swb; g++) {
if (w*16+g < sfb_start || w*16+g > sfb_end)
@@ -149,22 +171,26 @@ void ff_aac_search_for_tns(AACEncContext *s, SingleCha
else
en[0] += band->energy;
threshold += band->threshold;
- spread += band->spread;
+ spread += band->spread;
}
}
if (coef_len <= 0 || (sfb_end - sfb_start) <= 0)
continue;
- else
- e_ratio = en[0]/en[1];
/* LPC */
gain = ff_lpc_calc_ref_coefs_f(&s->lpc, &sce->coeffs[coef_start],
coef_len, order, coefs);
- if (gain > TNS_GAIN_THRESHOLD_LOW && gain < TNS_GAIN_THRESHOLD_HIGH &&
- (en[0]+en[1]) > TNS_GAIN_THRESHOLD_LOW*threshold &&
- spread < TNS_SPREAD_THRESHOLD && order) {
+ if (!order || gain < TNS_GAIN_THRESHOLD_LOW || gain > TNS_GAIN_THRESHOLD_HIGH)
+ use_tns = 0;
+ else if ((en[0]+en[1]) < TNS_GAIN_THRESHOLD_LOW*threshold || spread < TNS_SPREAD_THRESHOLD)
+ use_tns = 0;
+ else
+ use_tns = 1;
+
+ if (use_tns) {
+ e_ratio = en[0]/en[1];
if (is8 || order < 2 || (e_ratio > TNS_E_RATIO_LOW && e_ratio < TNS_E_RATIO_HIGH)) {
tns->n_filt[w] = 1;
for (g = 0; g < tns->n_filt[w]; g++) {
@@ -172,7 +198,7 @@ void ff_aac_search_for_tns(AACEncContext *s, SingleCha
tns->direction[w][g] = en[0] < en[1];
tns->order[w][g] = order;
quantize_coefs(coefs, tns->coef_idx[w][g], tns->coef[w][g],
- order);
+ order, c_bits);
}
} else { /* 2 filters due to energy disbalance */
tns->n_filt[w] = 2;
@@ -183,12 +209,11 @@ void ff_aac_search_for_tns(AACEncContext *s, SingleCha
(sfb_end - sfb_start) - tns->length[w][g-1];
quantize_coefs(&coefs[!g ? 0 : order - tns->order[w][g-1]],
tns->coef_idx[w][g], tns->coef[w][g],
- tns->order[w][g]);
+ tns->order[w][g], c_bits);
}
}
count++;
}
}
-
sce->tns.present = !!count;
}

32
graphics/ffmpeg/patches/patch-libavcodec_aacenc_tns_h Normal file
View File

@ -0,0 +1,32 @@
$OpenBSD: patch-libavcodec_aacenc_tns_h,v 1.1 2015/09/15 12:49:42 ajacoutot Exp $
aacenc_tns: redo coefficient quantization and decision making
aacenc_tns: readjust values for new TNS decision making
--- libavcodec/aacenc_tns.h.orig Sat Sep 12 15:01:35 2015
+++ libavcodec/aacenc_tns.h Sat Sep 12 15:02:13 2015
@@ -33,9 +33,12 @@
/* Could be set to 3 to save an additional bit at the cost of little quality */
#define TNS_Q_BITS 4
+/* Coefficient resolution in short windows */
+#define TNS_Q_BITS_SHORT 3
+
/* TNS will only be used if the LPC gain is within these margins */
-#define TNS_GAIN_THRESHOLD_LOW 1.395f
-#define TNS_GAIN_THRESHOLD_HIGH 11.19f
+#define TNS_GAIN_THRESHOLD_LOW 1.437f
+#define TNS_GAIN_THRESHOLD_HIGH 21.19f
/* If the energy ratio between the low SFBs vs the high SFBs is not between
* those two values, use 2 filters instead */
@@ -43,7 +46,7 @@
#define TNS_E_RATIO_HIGH 1.23
/* Do not use TNS if the psy band spread is below this value */
-#define TNS_SPREAD_THRESHOLD 37.081512f
+#define TNS_SPREAD_THRESHOLD 0.5f
void ff_aac_encode_tns_info(AACEncContext *s, SingleChannelElement *sce);
void ff_aac_apply_tns(AACEncContext *s, SingleChannelElement *sce);