falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
Browse Source

Cosmetics: Pretty print the AAC encoder. 

Originally committed as revision 19376 to svn://svn.ffmpeg.org/ffmpeg/trunk
tags/v0.6
Alex Converse 16 years ago
parent
78e65cd772
commit
fd257dc4c0
6 changed files with 288 additions and 288 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +169
    -169
      libavcodec/aaccoder.c
  2. +72
    -72
      libavcodec/aacenc.c
  3. +2
    -2
      libavcodec/aacenc.h
  4. +27
    -27
      libavcodec/aacpsy.c
  5. +11
    -11
      libavcodec/psymodel.c
  6. +7
    -7
      libavcodec/psymodel.h

+ 169
- 169
libavcodec/aaccoder.c View File

@@ -119,18 +119,18 @@ static float quantize_band_cost(struct AACEncContext *s, const float *in, const
int offs[4];
#endif /* USE_REALLY_FULL_SEARCH */

if(!cb){
for(i = 0; i < size; i++)
if (!cb) {
for (i = 0; i < size; i++)
cost += in[i]*in[i]*lambda;
return cost;
}
#ifndef USE_REALLY_FULL_SEARCH
offs[0] = 1;
for(i = 1; i < dim; i++)
for (i = 1; i < dim; i++)
offs[i] = offs[i-1]*range;
quantize_bands(s->qcoefs, in, scaled, size, Q34, !IS_CODEBOOK_UNSIGNED(cb), maxval);
#endif /* USE_REALLY_FULL_SEARCH */
for(i = 0; i < size; i += dim){
for (i = 0; i < size; i += dim) {
float mincost;
int minidx = 0;
int minbits = 0;
@@ -138,69 +138,69 @@ static float quantize_band_cost(struct AACEncContext *s, const float *in, const
#ifndef USE_REALLY_FULL_SEARCH
int (*quants)[2] = &s->qcoefs[i];
mincost = 0.0f;
for(j = 0; j < dim; j++){
for (j = 0; j < dim; j++) {
mincost += in[i+j]*in[i+j]*lambda;
}
minidx = IS_CODEBOOK_UNSIGNED(cb) ? 0 : 40;
minbits = ff_aac_spectral_bits[cb-1][minidx];
mincost += minbits;
for(j = 0; j < (1<<dim); j++){
for (j = 0; j < (1<<dim); j++) {
float rd = 0.0f;
int curbits;
int curidx = IS_CODEBOOK_UNSIGNED(cb) ? 0 : 40;
int same = 0;
for(k = 0; k < dim; k++){
if((j & (1 << k)) && quants[k][0] == quants[k][1]){
for (k = 0; k < dim; k++) {
if ((j & (1 << k)) && quants[k][0] == quants[k][1]) {
same = 1;
break;
}
}
if(same)
if (same)
continue;
for(k = 0; k < dim; k++)
for (k = 0; k < dim; k++)
curidx += quants[k][!!(j & (1 << k))] * offs[dim - 1 - k];
curbits = ff_aac_spectral_bits[cb-1][curidx];
vec = &ff_aac_codebook_vectors[cb-1][curidx*dim];
#else
mincost = INFINITY;
vec = ff_aac_codebook_vectors[cb-1];
for(j = 0; j < ff_aac_spectral_sizes[cb-1]; j++, vec += dim){
for (j = 0; j < ff_aac_spectral_sizes[cb-1]; j++, vec += dim) {
float rd = 0.0f;
int curbits = ff_aac_spectral_bits[cb-1][j];
#endif /* USE_REALLY_FULL_SEARCH */
if(IS_CODEBOOK_UNSIGNED(cb)){
for(k = 0; k < dim; k++){
if (IS_CODEBOOK_UNSIGNED(cb)) {
for (k = 0; k < dim; k++) {
float t = fabsf(in[i+k]);
float di;
//do not code with escape sequence small values
if(vec[k] == 64.0f && t < 39.0f*IQ){
if (vec[k] == 64.0f && t < 39.0f*IQ) {
rd = INFINITY;
break;
}
if(vec[k] == 64.0f){//FIXME: slow
if (vec[k] == 64.0f) {//FIXME: slow
if (t >= CLIPPED_ESCAPE) {
di = t - CLIPPED_ESCAPE;
curbits += 21;
}else{
} else {
int c = av_clip(quant(t, Q), 0, 8191);
di = t - c*cbrt(c)*IQ;
curbits += av_log2(c)*2 - 4 + 1;
}
}else{
} else {
di = t - vec[k]*IQ;
}
if(vec[k] != 0.0f)
if (vec[k] != 0.0f)
curbits++;
rd += di*di*lambda;
}
}else{
for(k = 0; k < dim; k++){
} else {
for (k = 0; k < dim; k++) {
float di = in[i+k] - vec[k]*IQ;
rd += di*di*lambda;
}
}
rd += curbits;
if(rd < mincost){
if (rd < mincost) {
mincost = rd;
minidx = j;
minbits = curbits;
@@ -208,11 +208,11 @@ static float quantize_band_cost(struct AACEncContext *s, const float *in, const
}
cost += mincost;
resbits += minbits;
if(cost >= uplim)
if (cost >= uplim)
return uplim;
}

if(bits)
if (bits)
*bits = resbits;
return cost;
}
@@ -234,17 +234,17 @@ static void quantize_and_encode_band(struct AACEncContext *s, PutBitContext *pb,
#endif /* USE_REALLY_FULL_SEARCH */

//START_TIMER
if(!cb)
if (!cb)
return;

#ifndef USE_REALLY_FULL_SEARCH
offs[0] = 1;
for(i = 1; i < dim; i++)
for (i = 1; i < dim; i++)
offs[i] = offs[i-1]*range;
abs_pow34_v(scaled, in, size);
quantize_bands(s->qcoefs, in, scaled, size, Q34, !IS_CODEBOOK_UNSIGNED(cb), maxval);
#endif /* USE_REALLY_FULL_SEARCH */
for(i = 0; i < size; i += dim){
for (i = 0; i < size; i += dim) {
float mincost;
int minidx = 0;
int minbits = 0;
@@ -252,83 +252,83 @@ static void quantize_and_encode_band(struct AACEncContext *s, PutBitContext *pb,
#ifndef USE_REALLY_FULL_SEARCH
int (*quants)[2] = &s->qcoefs[i];
mincost = 0.0f;
for(j = 0; j < dim; j++){
for (j = 0; j < dim; j++) {
mincost += in[i+j]*in[i+j]*lambda;
}
minidx = IS_CODEBOOK_UNSIGNED(cb) ? 0 : 40;
minbits = ff_aac_spectral_bits[cb-1][minidx];
mincost += minbits;
for(j = 0; j < (1<<dim); j++){
for (j = 0; j < (1<<dim); j++) {
float rd = 0.0f;
int curbits;
int curidx = IS_CODEBOOK_UNSIGNED(cb) ? 0 : 40;
int same = 0;
for(k = 0; k < dim; k++){
if((j & (1 << k)) && quants[k][0] == quants[k][1]){
for (k = 0; k < dim; k++) {
if ((j & (1 << k)) && quants[k][0] == quants[k][1]) {
same = 1;
break;
}
}
if(same)
if (same)
continue;
for(k = 0; k < dim; k++)
for (k = 0; k < dim; k++)
curidx += quants[k][!!(j & (1 << k))] * offs[dim - 1 - k];
curbits = ff_aac_spectral_bits[cb-1][curidx];
vec = &ff_aac_codebook_vectors[cb-1][curidx*dim];
#else
vec = ff_aac_codebook_vectors[cb-1];
mincost = INFINITY;
for(j = 0; j < ff_aac_spectral_sizes[cb-1]; j++, vec += dim){
for (j = 0; j < ff_aac_spectral_sizes[cb-1]; j++, vec += dim) {
float rd = 0.0f;
int curbits = ff_aac_spectral_bits[cb-1][j];
int curidx = j;
#endif /* USE_REALLY_FULL_SEARCH */
if(IS_CODEBOOK_UNSIGNED(cb)){
for(k = 0; k < dim; k++){
if (IS_CODEBOOK_UNSIGNED(cb)) {
for (k = 0; k < dim; k++) {
float t = fabsf(in[i+k]);
float di;
//do not code with escape sequence small values
if(vec[k] == 64.0f && t < 39.0f*IQ){
if (vec[k] == 64.0f && t < 39.0f*IQ) {
rd = INFINITY;
break;
}
if(vec[k] == 64.0f){//FIXME: slow
if (vec[k] == 64.0f) {//FIXME: slow
if (t >= CLIPPED_ESCAPE) {
di = t - CLIPPED_ESCAPE;
curbits += 21;
}else{
} else {
int c = av_clip(quant(t, Q), 0, 8191);
di = t - c*cbrt(c)*IQ;
curbits += av_log2(c)*2 - 4 + 1;
}
}else{
} else {
di = t - vec[k]*IQ;
}
if(vec[k] != 0.0f)
if (vec[k] != 0.0f)
curbits++;
rd += di*di*lambda;
}
}else{
for(k = 0; k < dim; k++){
} else {
for (k = 0; k < dim; k++) {
float di = in[i+k] - vec[k]*IQ;
rd += di*di*lambda;
}
}
rd += curbits;
if(rd < mincost){
if (rd < mincost) {
mincost = rd;
minidx = curidx;
minbits = curbits;
}
}
put_bits(pb, ff_aac_spectral_bits[cb-1][minidx], ff_aac_spectral_codes[cb-1][minidx]);
if(IS_CODEBOOK_UNSIGNED(cb))
for(j = 0; j < dim; j++)
if(ff_aac_codebook_vectors[cb-1][minidx*dim+j] != 0.0f)
if (IS_CODEBOOK_UNSIGNED(cb))
for (j = 0; j < dim; j++)
if (ff_aac_codebook_vectors[cb-1][minidx*dim+j] != 0.0f)
put_bits(pb, 1, in[i+j] < 0.0f);
if(cb == ESC_BT){
for(j = 0; j < 2; j++){
if(ff_aac_codebook_vectors[cb-1][minidx*2+j] == 64.0f){
if (cb == ESC_BT) {
for (j = 0; j < 2; j++) {
if (ff_aac_codebook_vectors[cb-1][minidx*2+j] == 64.0f) {
int coef = av_clip(quant(fabsf(in[i+j]), Q), 0, 8191);
int len = av_log2(coef);

@@ -370,29 +370,29 @@ static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce

abs_pow34_v(s->scoefs, sce->coeffs, 1024);
start = win*128;
for(cb = 0; cb < 12; cb++){
for (cb = 0; cb < 12; cb++) {
path[0][cb].cost = 0.0f;
path[0][cb].prev_idx = -1;
path[0][cb].run = 0;
}
for(swb = 0; swb < max_sfb; swb++){
for (swb = 0; swb < max_sfb; swb++) {
start2 = start;
size = sce->ics.swb_sizes[swb];
if(sce->zeroes[win*16 + swb]){
for(cb = 0; cb < 12; cb++){
if (sce->zeroes[win*16 + swb]) {
for (cb = 0; cb < 12; cb++) {
path[swb+1][cb].prev_idx = cb;
path[swb+1][cb].cost = path[swb][cb].cost;
path[swb+1][cb].run = path[swb][cb].run + 1;
}
}else{
} else {
float minrd = next_minrd;
int mincb = next_mincb;
next_minrd = INFINITY;
next_mincb = 0;
for(cb = 0; cb < 12; cb++){
for (cb = 0; cb < 12; cb++) {
float cost_stay_here, cost_get_here;
float rd = 0.0f;
for(w = 0; w < group_len; w++){
for (w = 0; w < group_len; w++) {
FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(win+w)*16+swb];
rd += quantize_band_cost(s, sce->coeffs + start + w*128,
s->scoefs + start + w*128, size,
@@ -401,7 +401,7 @@ static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce
}
cost_stay_here = path[swb][cb].cost + rd;
cost_get_here = minrd + rd + run_bits + 4;
if( run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run]
if ( run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run]
!= run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run+1])
cost_stay_here += run_bits;
if (cost_get_here < cost_stay_here) {
@@ -425,12 +425,12 @@ static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce
//convert resulting path from backward-linked list
stack_len = 0;
idx = 0;
for(cb = 1; cb < 12; cb++){
if(path[max_sfb][cb].cost < path[max_sfb][idx].cost)
for (cb = 1; cb < 12; cb++) {
if (path[max_sfb][cb].cost < path[max_sfb][idx].cost)
idx = cb;
}
ppos = max_sfb;
while(ppos > 0){
while(ppos > 0) {
cb = idx;
stackrun[stack_len] = path[ppos][cb].run;
stackcb [stack_len] = cb;
@@ -440,16 +440,16 @@ static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce
}
//perform actual band info encoding
start = 0;
for(i = stack_len - 1; i >= 0; i--){
for (i = stack_len - 1; i >= 0; i--) {
put_bits(&s->pb, 4, stackcb[i]);
count = stackrun[i];
memset(sce->zeroes + win*16 + start, !stackcb[i], count);
//XXX: memset when band_type is also uint8_t
for(j = 0; j < count; j++){
for (j = 0; j < count; j++) {
sce->band_type[win*16 + start] = stackcb[i];
start++;
}
while(count >= run_esc){
while(count >= run_esc) {
put_bits(&s->pb, run_bits, run_esc);
count -= run_esc;
}
@@ -482,13 +482,13 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
int minq;
float mincost;

for(i = 0; i < 256; i++){
for (i = 0; i < 256; i++) {
paths[i].cost = 0.0f;
paths[i].prev = -1;
paths[i].min_val = i;
paths[i].max_val = i;
}
for(i = 256; i < 256*121; i++){
for (i = 256; i < 256*121; i++) {
paths[i].cost = INFINITY;
paths[i].prev = -2;
paths[i].min_val = INT_MAX;
@@ -496,9 +496,9 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
}
idx = 256;
abs_pow34_v(s->scoefs, sce->coeffs, 1024);
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for(g = 0; g < sce->ics.num_swb; g++){
for (g = 0; g < sce->ics.num_swb; g++) {
const float *coefs = sce->coeffs + start;
float qmin, qmax;
int nz = 0;
@@ -506,53 +506,53 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
bandaddr[idx >> 8] = w*16+g;
qmin = INT_MAX;
qmax = 0.0f;
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
if(band->energy <= band->threshold || band->threshold == 0.0f){
if (band->energy <= band->threshold || band->threshold == 0.0f) {
sce->zeroes[(w+w2)*16+g] = 1;
continue;
}
sce->zeroes[(w+w2)*16+g] = 0;
nz = 1;
for(i = 0; i < sce->ics.swb_sizes[g]; i++){
for (i = 0; i < sce->ics.swb_sizes[g]; i++) {
float t = fabsf(coefs[w2*128+i]);
if(t > 0.0f) qmin = fminf(qmin, t);
if (t > 0.0f) qmin = fminf(qmin, t);
qmax = fmaxf(qmax, t);
}
}
if(nz){
if (nz) {
int minscale, maxscale;
float minrd = INFINITY;
//minimum scalefactor index is when minimum nonzero coefficient after quantizing is not clipped
minscale = av_clip_uint8(log2(qmin)*4 - 69 + SCALE_ONE_POS - SCALE_DIV_512);
//maximum scalefactor index is when maximum coefficient after quantizing is still not zero
maxscale = av_clip_uint8(log2(qmax)*4 + 6 + SCALE_ONE_POS - SCALE_DIV_512);
for(q = minscale; q < maxscale; q++){
for (q = minscale; q < maxscale; q++) {
float dists[12], dist;
memset(dists, 0, sizeof(dists));
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
int cb;
for(cb = 0; cb <= ESC_BT; cb++){
for (cb = 0; cb <= ESC_BT; cb++) {
dists[cb] += quantize_band_cost(s, coefs + w2*128, s->scoefs + start + w2*128, sce->ics.swb_sizes[g],
q, cb, lambda / band->threshold, INFINITY, NULL);
}
}
dist = dists[0];
for(i = 1; i <= ESC_BT; i++)
for (i = 1; i <= ESC_BT; i++)
dist = fminf(dist, dists[i]);
minrd = fminf(minrd, dist);

for(i = FFMAX(q - SCALE_MAX_DIFF, 0); i < FFMIN(q + SCALE_MAX_DIFF, 256); i++){
for (i = FFMAX(q - SCALE_MAX_DIFF, 0); i < FFMIN(q + SCALE_MAX_DIFF, 256); i++) {
float cost;
int minv, maxv;
if(isinf(paths[idx - 256 + i].cost))
if (isinf(paths[idx - 256 + i].cost))
continue;
cost = paths[idx - 256 + i].cost + dist
+ ff_aac_scalefactor_bits[q - i + SCALE_DIFF_ZERO];
minv = FFMIN(paths[idx - 256 + i].min_val, q);
maxv = FFMAX(paths[idx - 256 + i].max_val, q);
if(cost < paths[idx + q].cost && maxv-minv < SCALE_MAX_DIFF){
if (cost < paths[idx + q].cost && maxv-minv < SCALE_MAX_DIFF) {
paths[idx + q].cost = cost;
paths[idx + q].prev = idx - 256 + i;
paths[idx + q].min_val = minv;
@@ -560,24 +560,24 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
}
}
}
}else{
for(q = 0; q < 256; q++){
if(!isinf(paths[idx - 256 + q].cost)){
} else {
for (q = 0; q < 256; q++) {
if (!isinf(paths[idx - 256 + q].cost)) {
paths[idx + q].cost = paths[idx - 256 + q].cost + 1;
paths[idx + q].prev = idx - 256 + q;
paths[idx + q].min_val = FFMIN(paths[idx - 256 + q].min_val, q);
paths[idx + q].max_val = FFMAX(paths[idx - 256 + q].max_val, q);
continue;
}
for(i = FFMAX(q - SCALE_MAX_DIFF, 0); i < FFMIN(q + SCALE_MAX_DIFF, 256); i++){
for (i = FFMAX(q - SCALE_MAX_DIFF, 0); i < FFMIN(q + SCALE_MAX_DIFF, 256); i++) {
float cost;
int minv, maxv;
if(isinf(paths[idx - 256 + i].cost))
if (isinf(paths[idx - 256 + i].cost))
continue;
cost = paths[idx - 256 + i].cost + ff_aac_scalefactor_bits[q - i + SCALE_DIFF_ZERO];
minv = FFMIN(paths[idx - 256 + i].min_val, q);
maxv = FFMAX(paths[idx - 256 + i].max_val, q);
if(cost < paths[idx + q].cost && maxv-minv < SCALE_MAX_DIFF){
if (cost < paths[idx + q].cost && maxv-minv < SCALE_MAX_DIFF) {
paths[idx + q].cost = cost;
paths[idx + q].prev = idx - 256 + i;
paths[idx + q].min_val = minv;
@@ -594,20 +594,20 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
idx -= 256;
mincost = paths[idx].cost;
minq = idx;
for(i = 1; i < 256; i++){
if(paths[idx + i].cost < mincost){
for (i = 1; i < 256; i++) {
if (paths[idx + i].cost < mincost) {
mincost = paths[idx + i].cost;
minq = idx + i;
}
}
while(minq >= 256){
while(minq >= 256) {
sce->sf_idx[bandaddr[minq>>8]] = minq & 0xFF;
minq = paths[minq].prev;
}
//set the same quantizers inside window groups
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
for(g = 0; g < sce->ics.num_swb; g++)
for(w2 = 1; w2 < sce->ics.group_len[w]; w2++)
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
for (g = 0; g < sce->ics.num_swb; g++)
for (w2 = 1; w2 < sce->ics.group_len[w]; w2++)
sce->sf_idx[(w+w2)*16+g] = sce->sf_idx[w*16+g];
}

@@ -628,14 +628,14 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
//XXX: some heuristic to determine initial quantizers will reduce search time
memset(dists, 0, sizeof(dists));
//determine zero bands and upper limits
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for(g = 0; g < sce->ics.num_swb; g++){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
for (g = 0; g < sce->ics.num_swb; g++) {
int nz = 0;
float uplim = 0.0f;
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
uplim += band->threshold;
if(band->energy <= band->threshold || band->threshold == 0.0f){
if (band->energy <= band->threshold || band->threshold == 0.0f) {
sce->zeroes[(w+w2)*16+g] = 1;
continue;
}
@@ -643,14 +643,14 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
}
uplims[w*16+g] = uplim *512;
sce->zeroes[w*16+g] = !nz;
if(nz)
if (nz)
minthr = fminf(minthr, uplim);
allz = FFMAX(allz, nz);
}
}
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for(g = 0; g < sce->ics.num_swb; g++){
if(sce->zeroes[w*16+g]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
for (g = 0; g < sce->ics.num_swb; g++) {
if (sce->zeroes[w*16+g]) {
sce->sf_idx[w*16+g] = SCALE_ONE_POS;
continue;
}
@@ -658,7 +658,7 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
}
}

if(!allz)
if (!allz)
return;
abs_pow34_v(s->scoefs, sce->coeffs, 1024);
//perform two-loop search
@@ -672,9 +672,9 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
int prev = -1;
tbits = 0;
fflag = 0;
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for(g = 0; g < sce->ics.num_swb; g++){
for (g = 0; g < sce->ics.num_swb; g++) {
const float *coefs = sce->coeffs + start;
const float *scaled = s->scoefs + start;
int bits = 0;
@@ -682,13 +682,13 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
float mindist = INFINITY;
int minbits = 0;

if(sce->zeroes[w*16+g] || sce->sf_idx[w*16+g] >= 218)
if (sce->zeroes[w*16+g] || sce->sf_idx[w*16+g] >= 218)
continue;
minscaler = FFMIN(minscaler, sce->sf_idx[w*16+g]);
for(cb = 0; cb <= ESC_BT; cb++){
for (cb = 0; cb <= ESC_BT; cb++) {
float dist = 0.0f;
int bb = 0;
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
int b;
dist += quantize_band_cost(s, coefs + w2*128,
scaled + w2*128,
@@ -700,14 +700,14 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
&b);
bb += b;
}
if(dist < mindist){
if (dist < mindist) {
mindist = dist;
minbits = bb;
}
}
dists[w*16+g] = mindist - minbits;
bits = minbits;
if(prev != -1){
if (prev != -1) {
bits += ff_aac_scalefactor_bits[sce->sf_idx[w*16+g] - prev + SCALE_DIFF_ZERO];
}
tbits += bits;
@@ -715,36 +715,36 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, AACEncContext *
prev = sce->sf_idx[w*16+g];
}
}
if(tbits > destbits){
for(i = 0; i < 128; i++){
if(sce->sf_idx[i] < 218 - qstep){
if (tbits > destbits) {
for (i = 0; i < 128; i++) {
if (sce->sf_idx[i] < 218 - qstep) {
sce->sf_idx[i] += qstep;
}
}
}else{
for(i = 0; i < 128; i++){
if(sce->sf_idx[i] > 60 - qstep){
} else {
for (i = 0; i < 128; i++) {
if (sce->sf_idx[i] > 60 - qstep) {
sce->sf_idx[i] -= qstep;
}
}
}
qstep >>= 1;
if(!qstep && tbits > destbits*1.02)
if (!qstep && tbits > destbits*1.02)
qstep = 1;
if(sce->sf_idx[0] >= 217)break;
if (sce->sf_idx[0] >= 217)break;
}while(qstep);

fflag = 0;
minscaler = av_clip(minscaler, 60, 255 - SCALE_MAX_DIFF);
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for(g = 0; g < sce->ics.num_swb; g++){
for (g = 0; g < sce->ics.num_swb; g++) {
int prevsc = sce->sf_idx[w*16+g];
if(dists[w*16+g] > uplims[w*16+g] && sce->sf_idx[w*16+g] > 60)
if (dists[w*16+g] > uplims[w*16+g] && sce->sf_idx[w*16+g] > 60)
sce->sf_idx[w*16+g]--;
sce->sf_idx[w*16+g] = av_clip(sce->sf_idx[w*16+g], minscaler, minscaler + SCALE_MAX_DIFF);
sce->sf_idx[w*16+g] = FFMIN(sce->sf_idx[w*16+g], 219);
if(sce->sf_idx[w*16+g] != prevsc)
if (sce->sf_idx[w*16+g] != prevsc)
fflag = 1;
}
}
@@ -761,29 +761,29 @@ static void search_for_quantizers_faac(AVCodecContext *avctx, AACEncContext *s,
float distfact = ((sce->ics.num_windows > 1) ? 85.80 : 147.84) / lambda;
int last = 0, lastband = 0, curband = 0;
float avg_energy = 0.0;
if(sce->ics.num_windows == 1){
if (sce->ics.num_windows == 1) {
start = 0;
for(i = 0; i < 1024; i++){
if(i - start >= sce->ics.swb_sizes[curband]){
for (i = 0; i < 1024; i++) {
if (i - start >= sce->ics.swb_sizes[curband]) {
start += sce->ics.swb_sizes[curband];
curband++;
}
if(sce->coeffs[i]){
if (sce->coeffs[i]) {
avg_energy += sce->coeffs[i] * sce->coeffs[i];
last = i;
lastband = curband;
}
}
}else{
for(w = 0; w < 8; w++){
} else {
for (w = 0; w < 8; w++) {
const float *coeffs = sce->coeffs + w*128;
start = 0;
for(i = 0; i < 128; i++){
if(i - start >= sce->ics.swb_sizes[curband]){
for (i = 0; i < 128; i++) {
if (i - start >= sce->ics.swb_sizes[curband]) {
start += sce->ics.swb_sizes[curband];
curband++;
}
if(coeffs[i]){
if (coeffs[i]) {
avg_energy += coeffs[i] * coeffs[i];
last = FFMAX(last, i);
lastband = FFMAX(lastband, curband);
@@ -793,41 +793,41 @@ static void search_for_quantizers_faac(AVCodecContext *avctx, AACEncContext *s,
}
last++;
avg_energy /= last;
if(avg_energy == 0.0f){
for(i = 0; i < FF_ARRAY_ELEMS(sce->sf_idx); i++)
if (avg_energy == 0.0f) {
for (i = 0; i < FF_ARRAY_ELEMS(sce->sf_idx); i++)
sce->sf_idx[i] = SCALE_ONE_POS;
return;
}
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for(g = 0; g < sce->ics.num_swb; g++){
for (g = 0; g < sce->ics.num_swb; g++) {
float *coefs = sce->coeffs + start;
const int size = sce->ics.swb_sizes[g];
int start2 = start, end2 = start + size, peakpos = start;
float maxval = -1, thr = 0.0f, t;
maxq[w*16+g] = 0.0f;
if(g > lastband){
if (g > lastband) {
maxq[w*16+g] = 0.0f;
start += size;
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++)
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++)
memset(coefs + w2*128, 0, sizeof(coefs[0])*size);
continue;
}
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for(i = 0; i < size; i++){
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
for (i = 0; i < size; i++) {
float t = coefs[w2*128+i]*coefs[w2*128+i];
maxq[w*16+g] = fmaxf(maxq[w*16+g], fabsf(coefs[w2*128 + i]));
thr += t;
if(sce->ics.num_windows == 1 && maxval < t){
if (sce->ics.num_windows == 1 && maxval < t) {
maxval = t;
peakpos = start+i;
}
}
}
if(sce->ics.num_windows == 1){
if (sce->ics.num_windows == 1) {
start2 = FFMAX(peakpos - 2, start2);
end2 = FFMIN(peakpos + 3, end2);
}else{
} else {
start2 -= start;
end2 -= start;
}
@@ -839,16 +839,16 @@ static void search_for_quantizers_faac(AVCodecContext *avctx, AACEncContext *s,
}
memset(sce->sf_idx, 0, sizeof(sce->sf_idx));
abs_pow34_v(s->scoefs, sce->coeffs, 1024);
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for(g = 0; g < sce->ics.num_swb; g++){
for (g = 0; g < sce->ics.num_swb; g++) {
const float *coefs = sce->coeffs + start;
const float *scaled = s->scoefs + start;
const int size = sce->ics.swb_sizes[g];
int scf, prev_scf, step;
int min_scf = 0, max_scf = 255;
float curdiff;
if(maxq[w*16+g] < 21.544){
if (maxq[w*16+g] < 21.544) {
sce->zeroes[w*16+g] = 1;
start += size;
continue;
@@ -856,11 +856,11 @@ static void search_for_quantizers_faac(AVCodecContext *avctx, AACEncContext *s,
sce->zeroes[w*16+g] = 0;
scf = prev_scf = av_clip(SCALE_ONE_POS - SCALE_DIV_512 - log2(1/maxq[w*16+g])*16/3, 60, 218);
step = 16;
for(;;){
for (;;) {
float dist = 0.0f;
int quant_max;

for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
int b;
dist += quantize_band_cost(s, coefs + w2*128,
scaled + w2*128,
@@ -874,24 +874,24 @@ static void search_for_quantizers_faac(AVCodecContext *avctx, AACEncContext *s,
}
dist *= 1.0f/512.0f;
quant_max = quant(maxq[w*16+g], ff_aac_pow2sf_tab[200 - scf + SCALE_ONE_POS - SCALE_DIV_512]);
if(quant_max >= 8191){ // too much, return to the previous quantizer
if (quant_max >= 8191) { // too much, return to the previous quantizer
sce->sf_idx[w*16+g] = prev_scf;
break;
}
prev_scf = scf;
curdiff = fabsf(dist - uplim[w*16+g]);
if(curdiff == 0.0f)
if (curdiff == 0.0f)
step = 0;
else
step = fabsf(log2(curdiff));
if(dist > uplim[w*16+g])
if (dist > uplim[w*16+g])
step = -step;
if(FFABS(step) <= 1 || (step > 0 && scf >= max_scf) || (step < 0 && scf <= min_scf)){
if (FFABS(step) <= 1 || (step > 0 && scf >= max_scf) || (step < 0 && scf <= min_scf)) {
sce->sf_idx[w*16+g] = scf;
break;
}
scf += step;
if(step > 0)
if (step > 0)
min_scf = scf;
else
max_scf = scf;
@@ -900,17 +900,17 @@ static void search_for_quantizers_faac(AVCodecContext *avctx, AACEncContext *s,
}
}
minq = sce->sf_idx[0] ? sce->sf_idx[0] : INT_MAX;
for(i = 1; i < 128; i++){
if(!sce->sf_idx[i])
for (i = 1; i < 128; i++) {
if (!sce->sf_idx[i])
sce->sf_idx[i] = sce->sf_idx[i-1];
else
minq = FFMIN(minq, sce->sf_idx[i]);
}
if(minq == INT_MAX) minq = 0;
if (minq == INT_MAX) minq = 0;
minq = FFMIN(minq, SCALE_MAX_POS);
maxsf = FFMIN(minq + SCALE_MAX_DIFF, SCALE_MAX_POS);
for(i = 126; i >= 0; i--){
if(!sce->sf_idx[i])
for (i = 126; i >= 0; i--) {
if (!sce->sf_idx[i])
sce->sf_idx[i] = sce->sf_idx[i+1];
sce->sf_idx[i] = av_clip(sce->sf_idx[i], minq, maxsf);
}
@@ -923,15 +923,15 @@ static void search_for_quantizers_fast(AVCodecContext *avctx, AACEncContext *s,
int minq = 255;

memset(sce->sf_idx, 0, sizeof(sce->sf_idx));
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = w*128;
for(g = 0; g < sce->ics.num_swb; g++){
for(w2 = 0; w2 < sce->ics.group_len[w]; w2++){
for (g = 0; g < sce->ics.num_swb; g++) {
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
if(band->energy <= band->threshold){
if (band->energy <= band->threshold) {
sce->sf_idx[(w+w2)*16+g] = 218;
sce->zeroes[(w+w2)*16+g] = 1;
}else{
} else {
sce->sf_idx[(w+w2)*16+g] = av_clip(SCALE_ONE_POS - SCALE_DIV_512 + log2(band->threshold), 80, 218);
sce->zeroes[(w+w2)*16+g] = 0;
}
@@ -939,13 +939,13 @@ static void search_for_quantizers_fast(AVCodecContext *avctx, AACEncContext *s,
}
}
}
for(i = 0; i < 128; i++){
for (i = 0; i < 128; i++) {
sce->sf_idx[i] = 140;//av_clip(sce->sf_idx[i], minq, minq + SCALE_MAX_DIFF - 1);
}
//set the same quantizers inside window groups
for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
for(g = 0; g < sce->ics.num_swb; g++)
for(w2 = 1; w2 < sce->ics.group_len[w]; w2++)
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
for (g = 0; g < sce->ics.num_swb; g++)
for (w2 = 1; w2 < sce->ics.group_len[w]; w2++)
sce->sf_idx[(w+w2)*16+g] = sce->sf_idx[w*16+g];
}

@@ -956,18 +956,18 @@ static void search_for_ms(AACEncContext *s, ChannelElement *cpe, const float lam
float *L34 = s->scoefs, *R34 = s->scoefs + 128, *M34 = s->scoefs + 128*2, *S34 = s->scoefs + 128*3;
SingleChannelElement *sce0 = &cpe->ch[0];
SingleChannelElement *sce1 = &cpe->ch[1];
if(!cpe->common_window)
if (!cpe->common_window)
return;
for(w = 0; w < sce0->ics.num_windows; w += sce0->ics.group_len[w]){
for(g = 0; g < sce0->ics.num_swb; g++){
if(!cpe->ch[0].zeroes[w*16+g] && !cpe->ch[1].zeroes[w*16+g]){
for (w = 0; w < sce0->ics.num_windows; w += sce0->ics.group_len[w]) {
for (g = 0; g < sce0->ics.num_swb; g++) {
if (!cpe->ch[0].zeroes[w*16+g] && !cpe->ch[1].zeroes[w*16+g]) {
float dist1 = 0.0f, dist2 = 0.0f;
for(w2 = 0; w2 < sce0->ics.group_len[w]; w2++){
for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) {
FFPsyBand *band0 = &s->psy.psy_bands[(s->cur_channel+0)*PSY_MAX_BANDS+(w+w2)*16+g];
FFPsyBand *band1 = &s->psy.psy_bands[(s->cur_channel+1)*PSY_MAX_BANDS+(w+w2)*16+g];
float minthr = fminf(band0->threshold, band1->threshold);
float maxthr = fmaxf(band0->threshold, band1->threshold);
for(i = 0; i < sce0->ics.swb_sizes[g]; i++){
for (i = 0; i < sce0->ics.swb_sizes[g]; i++) {
M[i] = (sce0->coeffs[start+w2*128+i]
+ sce1->coeffs[start+w2*128+i])*0.5;
S[i] = sce0->coeffs[start+w2*128+i]


+ 72
- 72
libavcodec/aacenc.c View File

@@ -159,14 +159,14 @@ static av_cold int aac_encode_init(AVCodecContext *avctx)

avctx->frame_size = 1024;

for(i = 0; i < 16; i++)
if(avctx->sample_rate == ff_mpeg4audio_sample_rates[i])
for (i = 0; i < 16; i++)
if (avctx->sample_rate == ff_mpeg4audio_sample_rates[i])
break;
if(i == 16){
if (i == 16) {
av_log(avctx, AV_LOG_ERROR, "Unsupported sample rate %d\n", avctx->sample_rate);
return -1;
}
if(avctx->channels > 6){
if (avctx->channels > 6) {
av_log(avctx, AV_LOG_ERROR, "Unsupported number of channels: %d\n", avctx->channels);
return -1;
}
@@ -218,35 +218,35 @@ static void apply_window_and_mdct(AVCodecContext *avctx, AACEncContext *s,

if (sce->ics.window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
memcpy(s->output, sce->saved, sizeof(float)*1024);
if(sce->ics.window_sequence[0] == LONG_STOP_SEQUENCE){
if (sce->ics.window_sequence[0] == LONG_STOP_SEQUENCE) {
memset(s->output, 0, sizeof(s->output[0]) * 448);
for(i = 448; i < 576; i++)
for (i = 448; i < 576; i++)
s->output[i] = sce->saved[i] * pwindow[i - 448];
for(i = 576; i < 704; i++)
for (i = 576; i < 704; i++)
s->output[i] = sce->saved[i];
}
if(sce->ics.window_sequence[0] != LONG_START_SEQUENCE){
if (sce->ics.window_sequence[0] != LONG_START_SEQUENCE) {
j = channel;
for (i = 0; i < 1024; i++, j += avctx->channels){
for (i = 0; i < 1024; i++, j += avctx->channels) {
s->output[i+1024] = audio[j] * lwindow[1024 - i - 1];
sce->saved[i] = audio[j] * lwindow[i];
}
}else{
} else {
j = channel;
for(i = 0; i < 448; i++, j += avctx->channels)
for (i = 0; i < 448; i++, j += avctx->channels)
s->output[i+1024] = audio[j];
for(i = 448; i < 576; i++, j += avctx->channels)
for (i = 448; i < 576; i++, j += avctx->channels)
s->output[i+1024] = audio[j] * swindow[576 - i - 1];
memset(s->output+1024+576, 0, sizeof(s->output[0]) * 448);
j = channel;
for(i = 0; i < 1024; i++, j += avctx->channels)
for (i = 0; i < 1024; i++, j += avctx->channels)
sce->saved[i] = audio[j];
}
ff_mdct_calc(&s->mdct1024, sce->coeffs, s->output);
}else{
} else {
j = channel;
for (k = 0; k < 1024; k += 128) {
for(i = 448 + k; i < 448 + k + 256; i++)
for (i = 448 + k; i < 448 + k + 256; i++)
s->output[i - 448 - k] = (i < 1024)
? sce->saved[i]
: audio[channel + (i-1024)*avctx->channels];
@@ -255,7 +255,7 @@ static void apply_window_and_mdct(AVCodecContext *avctx, AACEncContext *s,
ff_mdct_calc(&s->mdct128, sce->coeffs + k, s->output);
}
j = channel;
for(i = 0; i < 1024; i++, j += avctx->channels)
for (i = 0; i < 1024; i++, j += avctx->channels)
sce->saved[i] = audio[j];
}
}
@@ -271,12 +271,12 @@ static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
put_bits(&s->pb, 1, 0); // ics_reserved bit
put_bits(&s->pb, 2, info->window_sequence[0]);
put_bits(&s->pb, 1, info->use_kb_window[0]);
if(info->window_sequence[0] != EIGHT_SHORT_SEQUENCE){
if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
put_bits(&s->pb, 6, info->max_sfb);
put_bits(&s->pb, 1, 0); // no prediction
}else{
} else {
put_bits(&s->pb, 4, info->max_sfb);
for(w = 1; w < 8; w++){
for (w = 1; w < 8; w++) {
put_bits(&s->pb, 1, !info->group_len[w]);
}
}
@@ -291,9 +291,9 @@ static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
int i, w;

put_bits(pb, 2, cpe->ms_mode);
if(cpe->ms_mode == 1){
for(w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w]){
for(i = 0; i < cpe->ch[0].ics.max_sfb; i++)
if (cpe->ms_mode == 1) {
for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w]) {
for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
}
}
@@ -307,34 +307,34 @@ static void adjust_frame_information(AACEncContext *apc, ChannelElement *cpe, in
int i, w, w2, g, ch;
int start, sum, maxsfb, cmaxsfb;

for(ch = 0; ch < chans; ch++){
for (ch = 0; ch < chans; ch++) {
IndividualChannelStream *ics = &cpe->ch[ch].ics;
start = 0;
maxsfb = 0;
cpe->ch[ch].pulse.num_pulse = 0;
for(w = 0; w < ics->num_windows*16; w += 16){
for(g = 0; g < ics->num_swb; g++){
for (w = 0; w < ics->num_windows*16; w += 16) {
for (g = 0; g < ics->num_swb; g++) {
sum = 0;
//apply M/S
if(!ch && cpe->ms_mask[w + g]){
for(i = 0; i < ics->swb_sizes[g]; i++){
if (!ch && cpe->ms_mask[w + g]) {
for (i = 0; i < ics->swb_sizes[g]; i++) {
cpe->ch[0].coeffs[start+i] = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) / 2.0;
cpe->ch[1].coeffs[start+i] = cpe->ch[0].coeffs[start+i] - cpe->ch[1].coeffs[start+i];
}
}
start += ics->swb_sizes[g];
}
for(cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w+cmaxsfb-1]; cmaxsfb--);
for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w+cmaxsfb-1]; cmaxsfb--);
maxsfb = FFMAX(maxsfb, cmaxsfb);
}
ics->max_sfb = maxsfb;

//adjust zero bands for window groups
for(w = 0; w < ics->num_windows; w += ics->group_len[w]){
for(g = 0; g < ics->max_sfb; g++){
for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
for (g = 0; g < ics->max_sfb; g++) {
i = 1;
for(w2 = w; w2 < w + ics->group_len[w]; w2++){
if(!cpe->ch[ch].zeroes[w2*16 + g]){
for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
if (!cpe->ch[ch].zeroes[w2*16 + g]) {
i = 0;
break;
}
@@ -344,16 +344,16 @@ static void adjust_frame_information(AACEncContext *apc, ChannelElement *cpe, in
}
}

if(chans > 1 && cpe->common_window){
if (chans > 1 && cpe->common_window) {
IndividualChannelStream *ics0 = &cpe->ch[0].ics;
IndividualChannelStream *ics1 = &cpe->ch[1].ics;
int msc = 0;
ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
ics1->max_sfb = ics0->max_sfb;
for(w = 0; w < ics0->num_windows*16; w += 16)
for(i = 0; i < ics0->max_sfb; i++)
if(cpe->ms_mask[w+i]) msc++;
if(msc == 0 || ics0->max_sfb == 0) cpe->ms_mode = 0;
for (w = 0; w < ics0->num_windows*16; w += 16)
for (i = 0; i < ics0->max_sfb; i++)
if (cpe->ms_mask[w+i]) msc++;
if (msc == 0 || ics0->max_sfb == 0) cpe->ms_mode = 0;
else cpe->ms_mode = msc < ics0->max_sfb ? 1 : 2;
}
}
@@ -365,7 +365,7 @@ static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
{
int w;

for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
}
}
@@ -378,11 +378,11 @@ static void encode_scale_factors(AVCodecContext *avctx, AACEncContext *s, Single
int off = sce->sf_idx[0], diff;
int i, w;

for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for(i = 0; i < sce->ics.max_sfb; i++){
if(!sce->zeroes[w*16 + i]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
for (i = 0; i < sce->ics.max_sfb; i++) {
if (!sce->zeroes[w*16 + i]) {
diff = sce->sf_idx[w*16 + i] - off + SCALE_DIFF_ZERO;
if(diff < 0 || diff > 120) av_log(avctx, AV_LOG_ERROR, "Scalefactor difference is too big to be coded\n");
if (diff < 0 || diff > 120) av_log(avctx, AV_LOG_ERROR, "Scalefactor difference is too big to be coded\n");
off = sce->sf_idx[w*16 + i];
put_bits(&s->pb, ff_aac_scalefactor_bits[diff], ff_aac_scalefactor_code[diff]);
}
@@ -398,11 +398,11 @@ static void encode_pulses(AACEncContext *s, Pulse *pulse)
int i;

put_bits(&s->pb, 1, !!pulse->num_pulse);
if(!pulse->num_pulse) return;
if (!pulse->num_pulse) return;

put_bits(&s->pb, 2, pulse->num_pulse - 1);
put_bits(&s->pb, 6, pulse->start);
for(i = 0; i < pulse->num_pulse; i++){
for (i = 0; i < pulse->num_pulse; i++) {
put_bits(&s->pb, 5, pulse->pos[i]);
put_bits(&s->pb, 4, pulse->amp[i]);
}
@@ -415,14 +415,14 @@ static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
{
int start, i, w, w2;

for(w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]){
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = 0;
for(i = 0; i < sce->ics.max_sfb; i++){
if(sce->zeroes[w*16 + i]){
for (i = 0; i < sce->ics.max_sfb; i++) {
if (sce->zeroes[w*16 + i]) {
start += sce->ics.swb_sizes[i];
continue;
}
for(w2 = w; w2 < w + sce->ics.group_len[w]; w2++){
for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++) {
s->coder->quantize_and_encode_band(s, &s->pb, sce->coeffs + start + w2*128,
sce->ics.swb_sizes[i],
sce->sf_idx[w*16 + i],
@@ -440,7 +440,7 @@ static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce, int common_window)
{
put_bits(&s->pb, 8, sce->sf_idx[0]);
if(!common_window) put_ics_info(s, &sce->ics);
if (!common_window) put_ics_info(s, &sce->ics);
encode_band_info(s, sce);
encode_scale_factors(avctx, s, sce);
encode_pulses(s, &sce->pulse);
@@ -460,12 +460,12 @@ static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s, const ch
namelen = strlen(name) + 2;
put_bits(&s->pb, 3, TYPE_FIL);
put_bits(&s->pb, 4, FFMIN(namelen, 15));
if(namelen >= 15)
if (namelen >= 15)
put_bits(&s->pb, 8, namelen - 16);
put_bits(&s->pb, 4, 0); //extension type - filler
padbits = 8 - (put_bits_count(&s->pb) & 7);
align_put_bits(&s->pb);
for(i = 0; i < namelen - 2; i++)
for (i = 0; i < namelen - 2; i++)
put_bits(&s->pb, 8, name[i]);
put_bits(&s->pb, 12 - padbits, 0);
}
@@ -480,15 +480,15 @@ static int aac_encode_frame(AVCodecContext *avctx,
const uint8_t *chan_map = aac_chan_configs[avctx->channels-1];
int chan_el_counter[4];

if(s->last_frame)
if (s->last_frame)
return 0;
if(data){
if(!s->psypp){
if (data) {
if (!s->psypp) {
memcpy(s->samples + 1024 * avctx->channels, data, 1024 * avctx->channels * sizeof(s->samples[0]));
}else{
} else {
start_ch = 0;
samples2 = s->samples + 1024 * avctx->channels;
for(i = 0; i < chan_map[0]; i++){
for (i = 0; i < chan_map[0]; i++) {
tag = chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
ff_psy_preprocess(s->psypp, (uint16_t*)data + start_ch, samples2 + start_ch, start_ch, chans);
@@ -496,26 +496,26 @@ static int aac_encode_frame(AVCodecContext *avctx,
}
}
}
if(!avctx->frame_number){
if (!avctx->frame_number) {
memcpy(s->samples, s->samples + 1024 * avctx->channels, 1024 * avctx->channels * sizeof(s->samples[0]));
return 0;
}

init_put_bits(&s->pb, frame, buf_size*8);
if((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT)){
if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT)) {
put_bitstream_info(avctx, s, LIBAVCODEC_IDENT);
}
start_ch = 0;
memset(chan_el_counter, 0, sizeof(chan_el_counter));
for(i = 0; i < chan_map[0]; i++){
for (i = 0; i < chan_map[0]; i++) {
FFPsyWindowInfo wi[2];
tag = chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
samples2 = samples + start_ch;
la = samples2 + 1024 * avctx->channels + start_ch;
if(!data) la = NULL;
for(j = 0; j < chans; j++){
if (!data) la = NULL;
for (j = 0; j < chans; j++) {
IndividualChannelStream *ics = &cpe->ch[j].ics;
int k;
wi[j] = ff_psy_suggest_window(&s->psy, samples2, la, start_ch + j, ics->window_sequence[0]);
@@ -526,7 +526,7 @@ static int aac_encode_frame(AVCodecContext *avctx,
ics->num_windows = wi[j].num_windows;
ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
ics->num_swb = s->psy.num_bands[ics->num_windows == 8];
for(k = 0; k < ics->num_windows; k++)
for (k = 0; k < ics->num_windows; k++)
ics->group_len[k] = wi[j].grouping[k];

s->cur_channel = start_ch + j;
@@ -534,31 +534,31 @@ static int aac_encode_frame(AVCodecContext *avctx,
s->coder->search_for_quantizers(avctx, s, &cpe->ch[j], s->lambda);
}
cpe->common_window = 0;
if(chans > 1
if (chans > 1
&& wi[0].window_type[0] == wi[1].window_type[0]
&& wi[0].window_shape == wi[1].window_shape){
&& wi[0].window_shape == wi[1].window_shape) {

cpe->common_window = 1;
for(j = 0; j < wi[0].num_windows; j++){
if(wi[0].grouping[j] != wi[1].grouping[j]){
for (j = 0; j < wi[0].num_windows; j++) {
if (wi[0].grouping[j] != wi[1].grouping[j]) {
cpe->common_window = 0;
break;
}
}
}
if(cpe->common_window && s->coder->search_for_ms)
if (cpe->common_window && s->coder->search_for_ms)
s->coder->search_for_ms(s, cpe, s->lambda);
adjust_frame_information(s, cpe, chans);
put_bits(&s->pb, 3, tag);
put_bits(&s->pb, 4, chan_el_counter[tag]++);
if(chans == 2){
if (chans == 2) {
put_bits(&s->pb, 1, cpe->common_window);
if(cpe->common_window){
if (cpe->common_window) {
put_ics_info(s, &cpe->ch[0].ics);
encode_ms_info(&s->pb, cpe);
}
}
for(j = 0; j < chans; j++){
for (j = 0; j < chans; j++) {
s->cur_channel = start_ch + j;
ff_psy_set_band_info(&s->psy, s->cur_channel, cpe->ch[j].coeffs, &wi[j]);
encode_individual_channel(avctx, s, &cpe->ch[j], cpe->common_window);
@@ -571,7 +571,7 @@ static int aac_encode_frame(AVCodecContext *avctx,
avctx->frame_bits = put_bits_count(&s->pb);

// rate control stuff
if(!(avctx->flags & CODEC_FLAG_QSCALE)){
if (!(avctx->flags & CODEC_FLAG_QSCALE)) {
float ratio = avctx->bit_rate * 1024.0f / avctx->sample_rate / avctx->frame_bits;
s->lambda *= ratio;
}
@@ -580,7 +580,7 @@ static int aac_encode_frame(AVCodecContext *avctx,
av_log(avctx, AV_LOG_ERROR, "input buffer violation %d > %d.\n", avctx->frame_bits, 6144*avctx->channels);
}

if(!data)
if (!data)
s->last_frame = 1;
memcpy(s->samples, s->samples + 1024 * avctx->channels, 1024 * avctx->channels * sizeof(s->samples[0]));
return put_bits_count(&s->pb)>>3;


+ 2
- 2
libavcodec/aacenc.h View File

@@ -32,7 +32,7 @@

struct AACEncContext;

typedef struct AACCoefficientsEncoder{
typedef struct AACCoefficientsEncoder {
void (*search_for_quantizers)(AVCodecContext *avctx, struct AACEncContext *s,
SingleChannelElement *sce, const float lambda);
void (*encode_window_bands_info)(struct AACEncContext *s, SingleChannelElement *sce,
@@ -40,7 +40,7 @@ typedef struct AACCoefficientsEncoder{
void (*quantize_and_encode_band)(struct AACEncContext *s, PutBitContext *pb, const float *in, int size,
int scale_idx, int cb, const float lambda);
void (*search_for_ms)(struct AACEncContext *s, ChannelElement *cpe, const float lambda);
}AACCoefficientsEncoder;
} AACCoefficientsEncoder;

extern AACCoefficientsEncoder ff_aac_coders[];



+ 27
- 27
libavcodec/aacpsy.c View File

@@ -112,7 +112,7 @@ static av_cold float ath(float f, float add)
+ (0.6 + 0.04 * add) * 0.001 * f * f * f * f;
}

static av_cold int psy_3gpp_init(FFPsyContext *ctx){
static av_cold int psy_3gpp_init(FFPsyContext *ctx) {
Psy3gppContext *pctx;
float barks[1024];
int i, j, g, start;
@@ -121,26 +121,26 @@ static av_cold int psy_3gpp_init(FFPsyContext *ctx){
ctx->model_priv_data = av_mallocz(sizeof(Psy3gppContext));
pctx = (Psy3gppContext*) ctx->model_priv_data;

for(i = 0; i < 1024; i++)
for (i = 0; i < 1024; i++)
barks[i] = calc_bark(i * ctx->avctx->sample_rate / 2048.0);
minath = ath(3410, ATH_ADD);
for(j = 0; j < 2; j++){
for (j = 0; j < 2; j++) {
Psy3gppCoeffs *coeffs = &pctx->psy_coef[j];
i = 0;
prev = 0.0;
for(g = 0; g < ctx->num_bands[j]; g++){
for (g = 0; g < ctx->num_bands[j]; g++) {
i += ctx->bands[j][g];
coeffs->barks[g] = (barks[i - 1] + prev) / 2.0;
prev = barks[i - 1];
}
for(g = 0; g < ctx->num_bands[j] - 1; g++){
for (g = 0; g < ctx->num_bands[j] - 1; g++) {
coeffs->spread_low[g] = pow(10.0, -(coeffs->barks[g+1] - coeffs->barks[g]) * PSY_3GPP_SPREAD_LOW);
coeffs->spread_hi [g] = pow(10.0, -(coeffs->barks[g+1] - coeffs->barks[g]) * PSY_3GPP_SPREAD_HI);
}
start = 0;
for(g = 0; g < ctx->num_bands[j]; g++){
for (g = 0; g < ctx->num_bands[j]; g++) {
minscale = ath(ctx->avctx->sample_rate * start / 1024.0, ATH_ADD);
for(i = 1; i < ctx->bands[j][g]; i++){
for (i = 1; i < ctx->bands[j][g]; i++) {
minscale = fminf(minscale, ath(ctx->avctx->sample_rate * (start + i) / 1024.0 / 2.0, ATH_ADD));
}
coeffs->ath[g] = minscale - minath;
@@ -189,21 +189,21 @@ static FFPsyWindowInfo psy_3gpp_window(FFPsyContext *ctx,
FFPsyWindowInfo wi;

memset(&wi, 0, sizeof(wi));
if(la){
if (la) {
float s[8], v;
int switch_to_eight = 0;
float sum = 0.0, sum2 = 0.0;
int attack_n = 0;
for(i = 0; i < 8; i++){
for(j = 0; j < 128; j++){
for (i = 0; i < 8; i++) {
for (j = 0; j < 128; j++) {
v = iir_filter(audio[(i*128+j)*ctx->avctx->channels], pch->iir_state);
sum += v*v;
}
s[i] = sum;
sum2 += sum;
}
for(i = 0; i < 8; i++){
if(s[i] > pch->win_energy * attack_ratio){
for (i = 0; i < 8; i++) {
if (s[i] > pch->win_energy * attack_ratio) {
attack_n = i + 1;
switch_to_eight = 1;
break;
@@ -212,7 +212,7 @@ static FFPsyWindowInfo psy_3gpp_window(FFPsyContext *ctx,
pch->win_energy = pch->win_energy*7/8 + sum2/64;

wi.window_type[1] = prev_type;
switch(prev_type){
switch (prev_type) {
case ONLY_LONG_SEQUENCE:
wi.window_type[0] = switch_to_eight ? LONG_START_SEQUENCE : ONLY_LONG_SEQUENCE;
break;
@@ -229,21 +229,21 @@ static FFPsyWindowInfo psy_3gpp_window(FFPsyContext *ctx,
break;
}
pch->next_grouping = window_grouping[attack_n];
}else{
for(i = 0; i < 3; i++)
} else {
for (i = 0; i < 3; i++)
wi.window_type[i] = prev_type;
grouping = (prev_type == EIGHT_SHORT_SEQUENCE) ? window_grouping[0] : 0;
}

wi.window_shape = 1;
if(wi.window_type[0] != EIGHT_SHORT_SEQUENCE){
if (wi.window_type[0] != EIGHT_SHORT_SEQUENCE) {
wi.num_windows = 1;
wi.grouping[0] = 1;
}else{
} else {
int lastgrp = 0;
wi.num_windows = 8;
for(i = 0; i < 8; i++){
if(!((grouping >> i) & 1))
for (i = 0; i < 8; i++) {
if (!((grouping >> i) & 1))
lastgrp = i;
wi.grouping[lastgrp]++;
}
@@ -267,11 +267,11 @@ static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, const float *coefs,
Psy3gppCoeffs *coeffs = &pctx->psy_coef[wi->num_windows == 8];

//calculate energies, initial thresholds and related values - 5.4.2 "Threshold Calculation"
for(w = 0; w < wi->num_windows*16; w += 16){
for(g = 0; g < num_bands; g++){
for (w = 0; w < wi->num_windows*16; w += 16) {
for (g = 0; g < num_bands; g++) {
Psy3gppBand *band = &pch->band[w+g];
band->energy = 0.0f;
for(i = 0; i < band_sizes[g]; i++)
for (i = 0; i < band_sizes[g]; i++)
band->energy += coefs[start+i] * coefs[start+i];
band->energy *= 1.0f / (512*512);
band->thr = band->energy * 0.001258925f;
@@ -281,17 +281,17 @@ static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, const float *coefs,
}
}
//modify thresholds - spread, threshold in quiet - 5.4.3 "Spreaded Energy Calculation"
for(w = 0; w < wi->num_windows*16; w += 16){
for (w = 0; w < wi->num_windows*16; w += 16) {
Psy3gppBand *band = &pch->band[w];
for(g = 1; g < num_bands; g++){
for (g = 1; g < num_bands; g++) {
band[g].thr = FFMAX(band[g].thr, band[g-1].thr * coeffs->spread_low[g-1]);
}
for(g = num_bands - 2; g >= 0; g--){
for (g = num_bands - 2; g >= 0; g--) {
band[g].thr = FFMAX(band[g].thr, band[g+1].thr * coeffs->spread_hi [g]);
}
for(g = 0; g < num_bands; g++){
for (g = 0; g < num_bands; g++) {
band[g].thr_quiet = FFMAX(band[g].thr, coeffs->ath[g]);
if(wi->num_windows != 8 && wi->window_type[1] != EIGHT_SHORT_SEQUENCE){
if (wi->num_windows != 8 && wi->window_type[1] != EIGHT_SHORT_SEQUENCE) {
band[g].thr_quiet = fmaxf(PSY_3GPP_RPEMIN*band[g].thr_quiet,
fminf(band[g].thr_quiet,
PSY_3GPP_RPELEV*pch->prev_band[w+g].thr_quiet));


+ 11
- 11
libavcodec/psymodel.c View File

@@ -35,12 +35,12 @@ av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx,
ctx->num_bands = av_malloc (sizeof(ctx->num_bands[0]) * num_lens);
memcpy(ctx->bands, bands, sizeof(ctx->bands[0]) * num_lens);
memcpy(ctx->num_bands, num_bands, sizeof(ctx->num_bands[0]) * num_lens);
switch(ctx->avctx->codec_id){
switch (ctx->avctx->codec_id) {
case CODEC_ID_AAC:
ctx->model = &ff_aac_psy_model;
break;
}
if(ctx->model->init)
if (ctx->model->init)
return ctx->model->init(ctx);
return 0;
}
@@ -60,7 +60,7 @@ void ff_psy_set_band_info(FFPsyContext *ctx, int channel,

av_cold void ff_psy_end(FFPsyContext *ctx)
{
if(ctx->model->end)
if (ctx->model->end)
ctx->model->end(ctx);
av_freep(&ctx->bands);
av_freep(&ctx->num_bands);
@@ -84,16 +84,16 @@ av_cold struct FFPsyPreprocessContext* ff_psy_preprocess_init(AVCodecContext *av
ctx = av_mallocz(sizeof(FFPsyPreprocessContext));
ctx->avctx = avctx;

if(avctx->flags & CODEC_FLAG_QSCALE)
if (avctx->flags & CODEC_FLAG_QSCALE)
cutoff_coeff = 1.0f / av_clip(1 + avctx->global_quality / FF_QUALITY_SCALE, 1, 8);
else
cutoff_coeff = avctx->bit_rate / (4.0f * avctx->sample_rate * avctx->channels);

ctx->fcoeffs = ff_iir_filter_init_coeffs(FF_FILTER_TYPE_BUTTERWORTH, FF_FILTER_MODE_LOWPASS,
FILT_ORDER, cutoff_coeff, 0.0, 0.0);
if(ctx->fcoeffs){
if (ctx->fcoeffs) {
ctx->fstate = av_mallocz(sizeof(ctx->fstate[0]) * avctx->channels);
for(i = 0; i < avctx->channels; i++)
for (i = 0; i < avctx->channels; i++)
ctx->fstate[i] = ff_iir_filter_init_state(FILT_ORDER);
}
return ctx;
@@ -104,15 +104,15 @@ void ff_psy_preprocess(struct FFPsyPreprocessContext *ctx,
int tag, int channels)
{
int ch, i;
if(ctx->fstate){
for(ch = 0; ch < channels; ch++){
if (ctx->fstate) {
for (ch = 0; ch < channels; ch++) {
ff_iir_filter(ctx->fcoeffs, ctx->fstate[tag+ch], ctx->avctx->frame_size,
audio + ch, ctx->avctx->channels,
dest + ch, ctx->avctx->channels);
}
}else{
for(ch = 0; ch < channels; ch++){
for(i = 0; i < ctx->avctx->frame_size; i++)
} else {
for (ch = 0; ch < channels; ch++) {
for (i = 0; i < ctx->avctx->frame_size; i++)
dest[i*ctx->avctx->channels + ch] = audio[i*ctx->avctx->channels + ch];
}
}


+ 7
- 7
libavcodec/psymodel.h View File

@@ -30,29 +30,29 @@
/**
* single band psychoacoustic information
*/
typedef struct FFPsyBand{
typedef struct FFPsyBand {
int bits;
float energy;
float threshold;
float distortion;
float perceptual_weight;
}FFPsyBand;
} FFPsyBand;

/**
* windowing related information
*/
typedef struct FFPsyWindowInfo{
typedef struct FFPsyWindowInfo {
int window_type[3]; ///< window type (short/long/transitional, etc.) - current, previous and next
int window_shape; ///< window shape (sine/KBD/whatever)
int num_windows; ///< number of windows in a frame
int grouping[8]; ///< window grouping (for e.g. AAC)
int *window_sizes; ///< sequence of window sizes inside one frame (for eg. WMA)
}FFPsyWindowInfo;
} FFPsyWindowInfo;

/**
* context used by psychoacoustic model
*/
typedef struct FFPsyContext{
typedef struct FFPsyContext {
AVCodecContext *avctx; ///< encoder context
const struct FFPsyModel *model; ///< encoder-specific model functions

@@ -63,7 +63,7 @@ typedef struct FFPsyContext{
int num_lens; ///< number of scalefactor band sets

void* model_priv_data; ///< psychoacoustic model implementation private data
}FFPsyContext;
} FFPsyContext;

/**
* codec-specific psychoacoustic model implementation
@@ -74,7 +74,7 @@ typedef struct FFPsyModel {
FFPsyWindowInfo (*window)(FFPsyContext *ctx, const int16_t *audio, const int16_t *la, int channel, int prev_type);
void (*analyze)(FFPsyContext *ctx, int channel, const float *coeffs, FFPsyWindowInfo *wi);
void (*end) (FFPsyContext *apc);
}FFPsyModel;
} FFPsyModel;

/**
* Initialize psychoacoustic model.


Write Preview
Loading…
Cancel
Save