[FFmpeg-cvslog] aacenc_tns: clamp filter direction energy measurement

[Lynne]

ffmpeg | branch: release/7.0 | Lynne <dev at lynne.ee> | Sat Feb  8 04:35:31 2025 +0100| [f98f142da571653436596ccad2d09c7e39bfd4fb] | committer: Lynne

aacenc_tns: clamp filter direction energy measurement

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

(cherry picked from commit ed09aa28ae3b4509f00a24a9ebdeb084ee00736a)

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=f98f142da571653436596ccad2d09c7e39bfd4fb
---

 libavcodec/aacenc_tns.c | 40 +++++++++++++++++++++++++++-------------
 1 file changed, 27 insertions(+), 13 deletions(-)

diff --git a/libavcodec/aacenc_tns.c b/libavcodec/aacenc_tns.c
index 60888fece7..f9bc033b26 100644
--- a/libavcodec/aacenc_tns.c
+++ b/libavcodec/aacenc_tns.c
@@ -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};
-        int oc_start = 0, os_start = 0;
+        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 */
@@ -199,15 +214,14 @@ 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->order[w][g] = g < tns->n_filt[w] ? order/tns->n_filt[w] : order - oc_start;
-            tns->length[w][g] = g < tns->n_filt[w] ? sfb_len/tns->n_filt[w] : sfb_len - os_start;
+            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],
                             tns->order[w][g], c_bits);
             oc_start += tns->order[w][g];
-            os_start += tns->length[w][g];
         }
         count++;
     }