shithub: opus

Info  •  Files  •  Log  •  Branches

Adds "temporal VBR", which should be mostly useful at low bitrate.

The idea is to allow more noise than the default in quiet segments.
Right now, this hurts for high bitrates , so it's gradually reduced
as the rate goes up, until it's completely disabled above 68 kb/s.

--- a/celt/celt_encoder.c

+++ b/celt/celt_encoder.c

@@ -113,6 +113,7 @@

    int intensity;

    opus_val16 *energy_save;

    opus_val16 *energy_mask;

+   opus_val16 spec_avg;

 #ifdef RESYNTH

    /* +MAX_PERIOD/2 to make space for overlap */

@@ -1115,7 +1116,8 @@

       int LM, opus_int32 bitrate, int lastCodedBands, int C, int intensity,

       int constrained_vbr, opus_val16 stereo_saving, int tot_boost,

       opus_val16 tf_estimate, int pitch_change, opus_val16 maxDepth,

-      int variable_duration, int lfe, int has_surround_mask, opus_val16 surround_masking)

+      int variable_duration, int lfe, int has_surround_mask, opus_val16 surround_masking,

+      opus_val16 temporal_vbr)

 {

    /* The target rate in 8th bits per frame */

    opus_int32 target;

@@ -1208,6 +1210,16 @@

       target = base_target + (opus_int32)MULT16_32_Q15(rate_factor, target-base_target);

    }

+

+   if (tf_estimate < QCONST16(.2f, 14))

+   {

+      opus_val16 amount;

+      opus_val16 tvbr_factor;

+      amount = MULT16_16_Q15(QCONST16(.000006f, 30), IMAX(0, IMIN(42000, 68000-bitrate)));

+      tvbr_factor = SHR32(MULT16_16(temporal_vbr, amount), DB_SHIFT);

+      target += (opus_int32)MULT16_32_Q15(tvbr_factor, target);

+   }

+

    /* Don't allow more than doubling the rate */

    target = IMIN(2*base_target, target);

@@ -1275,6 +1287,7 @@

    int signalBandwidth;

    int transient_got_disabled=0;

    opus_val16 surround_masking=0;

+   opus_val16 temporal_vbr=0;

    ALLOC_STACK;

    mode = st->mode;

@@ -1538,6 +1551,22 @@

       surround_masking = DIV32_16(mask_avg,C*st->end) + QCONST16(.0f, DB_SHIFT);

       surround_masking = MIN16(MAX16(surround_masking,-QCONST16(1.5f, DB_SHIFT)), 0);

    }

+   {

+      opus_val16 follow=-QCONST16(10.0f,DB_SHIFT);

+      float frame_avg=0;

+      opus_val16 offset = shortBlocks?HALF16(SHL16(LM, DB_SHIFT)):0;

+      for(i=st->start;i<st->end;i++)

+      {

+         follow = MAX16(follow-QCONST16(1.f, DB_SHIFT), bandLogE[i]-offset);

+         if (C==2)

+            follow = MAX16(follow, bandLogE[i+nbEBands]-offset);

+         frame_avg += follow;

+      }

+      frame_avg /= (st->end-st->start);

+      temporal_vbr = SUB16(frame_avg,st->spec_avg);

+      temporal_vbr = MIN16(QCONST16(3.f, DB_SHIFT), MAX16(-QCONST16(1.5f, DB_SHIFT), temporal_vbr));

+      st->spec_avg += MULT16_16_Q15(QCONST16(.02f, 15), temporal_vbr);

+   }

    /*for (i=0;i<21;i++)

       printf("%f ", bandLogE[i]);

    printf("\n");*/

@@ -1744,7 +1773,8 @@

      target = compute_vbr(mode, &st->analysis, base_target, LM, st->bitrate,

            st->lastCodedBands, C, st->intensity, st->constrained_vbr,

            st->stereo_saving, tot_boost, tf_estimate, pitch_change, maxDepth,

-           st->variable_duration, st->lfe, st->energy_mask!=NULL, surround_masking);

+           st->variable_duration, st->lfe, st->energy_mask!=NULL, surround_masking,

+           temporal_vbr);

      /* The current offset is removed from the target and the space used

         so far is added*/