wrapper/FFmpeg
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

aacenc_tns: clamp filter direction energy measurement

Browse source 
The issue is that:

float en[2];
...
tns->n_filt[w] = is8 ? 1 : order != TNS_MAX_ORDER ? 2 : 3;
for (g = 0; g < tns->n_filt[w]; g++) {
    tns->direction[w][g] = slant != 2 ? slant : en[g] < en[!g];

When using the AAC Main profile, n_filt = 3, and slant is by
default 2 (normal long frames), g can go above 1.

en is the evolution of energy in the frequency domain for every
band at the given window. E.g. whether the energy is concentrated
at the top of each band, or the bottom.

For 2-pole filters, its straightforward.
For 3-pole filters, we need more than 2 measurements.

This commit properly implements support for 3-pole filters, by measuring
the band energy across three areas.

Do note that even xHE-AAC caps n_filt to 2, and only AAC Main allows
n_filt == 3.

Fixes https://trac.ffmpeg.org/ticket/11418
This commit is contained in:
Lynne 2025-02-08 04:35:31 +01:00
parent e9cd280dfe
commit bedfb6eca4
No known key found for this signature in database
GPG key ID: A2FEA5F03F034464
1 changed files with 24 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

33
libavcodec/aacenc_tns.c
View file

@ -173,6 +173,7 @@ void ff_aac_search_for_tns(AACEncContext *s, SingleChannelElement *sce)
sce->ics.window_sequence[0] == LONG_START_SEQUENCE ? 0 : 2;
const int sfb_len = sfb_end - sfb_start;
const int coef_len = sce->ics.swb_offset[sfb_end] - sce->ics.swb_offset[sfb_start];
const int n_filt = is8 ? 1 : order != TNS_MAX_ORDER ? 2 : 3;
if (coef_len <= 0 || sfb_len <= 0) {
sce->tns.present = 0;
@ -180,16 +181,30 @@ void ff_aac_search_for_tns(AACEncContext *s, SingleChannelElement *sce)
}
for (w = 0; w < sce->ics.num_windows; w++) {
float en[2] = {0.0f, 0.0f};
float en[4] = {0.0f, 0.0f, 0.0f, 0.0f};
int oc_start = 0;
int coef_start = sce->ics.swb_offset[sfb_start];
for (g = sfb_start; g < sce->ics.num_swb && g <= sfb_end; g++) {
FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[w*16+g];
if (g > sfb_start + (sfb_len/2))
en[1] += band->energy;
else
en[0] += band->energy;
if (n_filt == 2) {
for (g = sfb_start; g < sce->ics.num_swb && g <= sfb_end; g++) {
FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[w*16+g];
if (g > sfb_start + (sfb_len/2))
en[1] += band->energy; /* End */
else
en[0] += band->energy; /* Start */
}
en[2] = en[0];
} else {
for (g = sfb_start; g < sce->ics.num_swb && g <= sfb_end; g++) {
FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[w*16+g];
if (g > sfb_start + (sfb_len/2) + (sfb_len/4))
en[2] += band->energy; /* End */
else if (g > sfb_start + (sfb_len/2) - (sfb_len/4))
en[1] += band->energy; /* Middle */
else
en[0] += band->energy; /* Start */
}
en[3] = en[0];
}
/* LPC */
@ -199,9 +214,9 @@ void ff_aac_search_for_tns(AACEncContext *s, SingleChannelElement *sce)
if (!order || !isfinite(gain) || gain < TNS_GAIN_THRESHOLD_LOW || gain > TNS_GAIN_THRESHOLD_HIGH)
continue;
tns->n_filt[w] = is8 ? 1 : order != TNS_MAX_ORDER ? 2 : 3;
tns->n_filt[w] = n_filt;
for (g = 0; g < tns->n_filt[w]; g++) {
tns->direction[w][g] = slant != 2 ? slant : en[g] < en[!g];
tns->direction[w][g] = slant != 2 ? slant : en[g] < en[g + 1];
tns->order[w][g] = order/tns->n_filt[w];
tns->length[w][g] = sfb_len/tns->n_filt[w];
quantize_coefs(&coefs[oc_start], tns->coef_idx[w][g], tns->coef[w][g],