From 63855aff731dcf58875c159dc2fa7463a444d617 Mon Sep 17 00:00:00 2001
From: Jean-Marc Valin <jmvalin@jmvalin.ca>
Date: Sun, 26 Jun 2022 12:48:44 -0400
Subject: Improve background noise estimation for CELT DTX

We now update the background noise estimate even in frames classified
as transient. It shouldn't be a problem because we're using min
statistics. Also, it avoids problems when update frames get
missclassified as transient.

In addition, we now use the duration of losses rather than the
number of lost packets to make decisions. That should make
PLC/DTX behaviour more consistent across frame sizes.
---
 celt/celt_decoder.c | 40 +++++++++++++++++++---------------------
 1 file changed, 19 insertions(+), 21 deletions(-)

(limited to 'celt')

diff --git a/celt/celt_decoder.c b/celt/celt_decoder.c
index 74ca3b74..77eb44f4 100644
--- a/celt/celt_decoder.c
+++ b/celt/celt_decoder.c
@@ -90,7 +90,7 @@ struct OpusCustomDecoder {
    opus_uint32 rng;
    int error;
    int last_pitch_index;
-   int loss_count;
+   int loss_duration;
    int skip_plc;
    int postfilter_period;
    int postfilter_period_old;
@@ -512,7 +512,7 @@ static void celt_decode_lost(CELTDecoder * OPUS_RESTRICT st, int N, int LM)
    int nbEBands;
    int overlap;
    int start;
-   int loss_count;
+   int loss_duration;
    int noise_based;
    const opus_int16 *eBands;
    SAVE_STACK;
@@ -532,9 +532,9 @@ static void celt_decode_lost(CELTDecoder * OPUS_RESTRICT st, int N, int LM)
    oldLogE2 = oldLogE + 2*nbEBands;
    backgroundLogE = oldLogE2  + 2*nbEBands;
 
-   loss_count = st->loss_count;
+   loss_duration = st->loss_duration;
    start = st->start;
-   noise_based = loss_count >= 5 || start != 0 || st->skip_plc;
+   noise_based = loss_duration >= 40 || start != 0 || st->skip_plc;
    if (noise_based)
    {
       /* Noise-based PLC/CNG */
@@ -559,7 +559,7 @@ static void celt_decode_lost(CELTDecoder * OPUS_RESTRICT st, int N, int LM)
 #endif
 
       /* Energy decay */
-      decay = loss_count==0 ? QCONST16(1.5f, DB_SHIFT) : QCONST16(.5f, DB_SHIFT);
+      decay = loss_duration==0 ? QCONST16(1.5f, DB_SHIFT) : QCONST16(.5f, DB_SHIFT);
       c=0; do
       {
          for (i=start;i<end;i++)
@@ -602,7 +602,7 @@ static void celt_decode_lost(CELTDecoder * OPUS_RESTRICT st, int N, int LM)
       VARDECL(opus_val16, _exc);
       VARDECL(opus_val16, fir_tmp);
 
-      if (loss_count == 0)
+      if (loss_duration == 0)
       {
          st->last_pitch_index = pitch_index = celt_plc_pitch_search(decode_mem, C, st->arch);
       } else {
@@ -632,7 +632,7 @@ static void celt_decode_lost(CELTDecoder * OPUS_RESTRICT st, int N, int LM)
          for (i=0;i<MAX_PERIOD+LPC_ORDER;i++)
             exc[i-LPC_ORDER] = ROUND16(buf[DECODE_BUFFER_SIZE-MAX_PERIOD-LPC_ORDER+i], SIG_SHIFT);
 
-         if (loss_count == 0)
+         if (loss_duration == 0)
          {
             opus_val32 ac[LPC_ORDER+1];
             /* Compute LPC coefficients for the last MAX_PERIOD samples before
@@ -812,7 +812,8 @@ static void celt_decode_lost(CELTDecoder * OPUS_RESTRICT st, int N, int LM)
       } while (++c<C);
    }
 
-   st->loss_count = loss_count+1;
+   /* Saturate to soemthing large to avoid wrap-around. */
+   st->loss_duration = IMIN(10000, loss_duration+(1<<LM));
 
    RESTORE_STACK;
 }
@@ -868,6 +869,7 @@ int celt_decode_with_ec(CELTDecoder * OPUS_RESTRICT st, const unsigned char *dat
    int nbEBands;
    int overlap;
    const opus_int16 *eBands;
+   opus_val16 max_background_increase;
    ALLOC_STACK;
 
    VALIDATE_CELT_DECODER(st);
@@ -942,7 +944,7 @@ int celt_decode_with_ec(CELTDecoder * OPUS_RESTRICT st, const unsigned char *dat
 
    /* Check if there are at least two packets received consecutively before
     * turning on the pitch-based PLC */
-   st->skip_plc = st->loss_count != 0;
+   st->skip_plc = st->loss_duration != 0;
 
    if (dec == NULL)
    {
@@ -1140,25 +1142,21 @@ int celt_decode_with_ec(CELTDecoder * OPUS_RESTRICT st, const unsigned char *dat
    if (C==1)
       OPUS_COPY(&oldBandE[nbEBands], oldBandE, nbEBands);
 
-   /* In case start or end were to change */
    if (!isTransient)
    {
-      opus_val16 max_background_increase;
       OPUS_COPY(oldLogE2, oldLogE, 2*nbEBands);
       OPUS_COPY(oldLogE, oldBandE, 2*nbEBands);
-      /* In normal circumstances, we only allow the noise floor to increase by
-         up to 2.4 dB/second, but when we're in DTX, we allow up to 6 dB
-         increase for each update.*/
-      if (st->loss_count < 10)
-         max_background_increase = M*QCONST16(0.001f,DB_SHIFT);
-      else
-         max_background_increase = QCONST16(1.f,DB_SHIFT);
-      for (i=0;i<2*nbEBands;i++)
-         backgroundLogE[i] = MIN16(backgroundLogE[i] + max_background_increase, oldBandE[i]);
    } else {
       for (i=0;i<2*nbEBands;i++)
          oldLogE[i] = MIN16(oldLogE[i], oldBandE[i]);
    }
+   /* In normal circumstances, we only allow the noise floor to increase by
+      up to 2.4 dB/second, but when we're in DTX we give the weight of
+      all missing packets to the update packet. */
+   max_background_increase = IMIN(160, st->loss_duration+M)*QCONST16(0.001f,DB_SHIFT);
+   for (i=0;i<2*nbEBands;i++)
+      backgroundLogE[i] = MIN16(backgroundLogE[i] + max_background_increase, oldBandE[i]);
+   /* In case start or end were to change */
    c=0; do
    {
       for (i=0;i<start;i++)
@@ -1175,7 +1173,7 @@ int celt_decode_with_ec(CELTDecoder * OPUS_RESTRICT st, const unsigned char *dat
    st->rng = dec->rng;
 
    deemphasis(out_syn, pcm, N, CC, st->downsample, mode->preemph, st->preemph_memD, accum);
-   st->loss_count = 0;
+   st->loss_duration = 0;
    RESTORE_STACK;
    if (ec_tell(dec) > 8*len)
       return OPUS_INTERNAL_ERROR;
-- 
cgit v1.2.3