Fixes to the the activity flag that is passed to Silk so it represents the final activity flag used in the DTX decision

This flag was modified after calling the Silk encoder function. This commit corrects that behavior by introducing those modifications before calling the Silk encoder.

Slightly modified comments by Felicia Lim

Signed-off-by: Felicia Lim <flim@google.com>

1 files changed, 19 insertions, 35 deletions

diff --git a/src/opus_encoder.c b/src/opus_encoder.c
index 844b08dd..1ae5598a 100644
--- a/src/opus_encoder.c
+++ b/src/opus_encoder.c
@@ -892,34 +892,15 @@ static opus_val32 compute_frame_energy(const opus_val16 *pcm, int frame_size, in
 #endif
 
 /* Decides if DTX should be turned on (=1) or off (=0) */
-static int decide_dtx_mode(float activity_probability,    /* probability that current frame contains speech/music */
-                           int *nb_no_activity_frames,    /* number of consecutive frames with no activity */
-                           opus_val32 peak_signal_energy, /* peak energy of desired signal detected so far */
-                           const opus_val16 *pcm,         /* input pcm signal */
-                           int frame_size,                /* frame size */
-                           int channels,
-                           int is_silence,                 /* only digital silence detected in this frame */
-                           int arch
-                          )
-{
-   opus_val32 noise_energy;
+static int decide_dtx_mode(opus_int activity,          /* indicates if this frame contains speech/music */
+                           int *nb_no_activity_frames  /* number of consecutive frames with no activity */
+                           )
 
-   if (!is_silence)
-   {
-      if (activity_probability < DTX_ACTIVITY_THRESHOLD)  /* is noise */
-      {
-         noise_energy = compute_frame_energy(pcm, frame_size, channels, arch);
-
-         /* but is sufficiently quiet */
-         is_silence = peak_signal_energy >= (PSEUDO_SNR_THRESHOLD * noise_energy);
-      }
-   }
-
-   if (is_silence)
+{
+   if (!activity)
    {
       /* The number of consecutive DTX frames should be within the allowed bounds */
       (*nb_no_activity_frames)++;
-
       if (*nb_no_activity_frames > NB_SPEECH_FRAMES_BEFORE_DTX)
       {
          if (*nb_no_activity_frames <= (NB_SPEECH_FRAMES_BEFORE_DTX + MAX_CONSECUTIVE_DTX))
@@ -1102,6 +1083,8 @@ opus_int32 opus_encode_native(OpusEncoder *st, const opus_val16 *pcm, int frame_
     int analysis_read_subframe_bak=-1;
     int is_silence = 0;
 #endif
+    opus_int activity = VAD_NO_DECISION;
+
     VARDECL(opus_val16, tmp_prefill);
 
     ALLOC_STACK;
@@ -1169,6 +1152,17 @@ opus_int32 opus_encode_native(OpusEncoder *st, const opus_val16 *pcm, int frame_
     if (!is_silence)
       st->voice_ratio = -1;
 
+    if (analysis_info.valid) {
+       activity = !is_silence && analysis_info.activity_probability >= DTX_ACTIVITY_THRESHOLD;
+       if (!activity) {
+          /* Mark as active if this noise frame is sufficiently loud */
+          opus_val32 noise_energy = compute_frame_energy(pcm, frame_size, st->channels, st->arch);
+          activity = st->peak_signal_energy < (PSEUDO_SNR_THRESHOLD * noise_energy);
+       }
+    } else {
+       activity = !is_silence;
+    }
+
     st->detected_bandwidth = 0;
     if (analysis_info.valid)
     {
@@ -1668,7 +1662,6 @@ opus_int32 opus_encode_native(OpusEncoder *st, const opus_val16 *pcm, int frame_
     if (st->mode != MODE_CELT_ONLY)
     {
         opus_int32 total_bitRate, celt_rate;
-        opus_int activity;
 #ifdef FIXED_POINT
        const opus_int16 *pcm_silk;
 #else
@@ -1676,14 +1669,6 @@ opus_int32 opus_encode_native(OpusEncoder *st, const opus_val16 *pcm, int frame_
        ALLOC(pcm_silk, st->channels*frame_size, opus_int16);
 #endif
 
-        activity = VAD_NO_DECISION;
-#ifndef DISABLE_FLOAT_API
-        if( analysis_info.valid ) {
-            /* Inform SILK about the Opus VAD decision */
-            activity = ( analysis_info.activity_probability >= DTX_ACTIVITY_THRESHOLD );
-        }
-#endif
-
         /* Distribute bits between SILK and CELT */
         total_bitRate = 8 * bytes_target * frame_rate;
         if( st->mode == MODE_HYBRID ) {
@@ -2144,8 +2129,7 @@ opus_int32 opus_encode_native(OpusEncoder *st, const opus_val16 *pcm, int frame_
 #ifndef DISABLE_FLOAT_API
     if (st->use_dtx && (analysis_info.valid || is_silence))
     {
-       if (decide_dtx_mode(analysis_info.activity_probability, &st->nb_no_activity_frames,
-             st->peak_signal_energy, pcm, frame_size, st->channels, is_silence, st->arch))
+       if (decide_dtx_mode(activity, &st->nb_no_activity_frames))
        {
           st->rangeFinal = 0;
           data[0] = gen_toc(st->mode, st->Fs/frame_size, curr_bandwidth, st->stream_channels);