Explicitly use the Opus DTX encoder state.

Use the DTX state from inside the Opus encoder instead of trying to
mimic the logic outside.

Bug: None
Change-Id: I852044fee261a5b7f9255c557a27adfd0b1701bd
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/362640
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Lionel Koenig <lionelk@webrtc.org>
Reviewed-by: Jakob Ivarsson‎ <jakobi@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#43034}

modules/audio_coding/codecs/opus/audio_encoder_opus.cc

@@ -378,8 +378,7 @@ AudioEncoderOpusImpl::AudioEncoderOpusImpl(
       inst_(nullptr),
       packet_loss_fraction_smoother_(new PacketLossFractionSmoother()),
       audio_network_adaptor_creator_(audio_network_adaptor_creator),
-      bitrate_smoother_(std::move(bitrate_smoother)),
+      bitrate_smoother_(std::move(bitrate_smoother)) {
-      consecutive_dtx_frames_(0) {
   RTC_DCHECK(0 <= payload_type && payload_type <= 127);
 
   // Sanity check of the redundant payload type field that we want to get rid
@@ -387,6 +386,7 @@ AudioEncoderOpusImpl::AudioEncoderOpusImpl(
   RTC_CHECK(config.payload_type == -1 || config.payload_type == payload_type);
 
   RTC_CHECK(RecreateEncoderInstance(config));
+
   SetProjectedPacketLossRate(packet_loss_rate_);
 }
 
@@ -609,8 +609,6 @@ AudioEncoder::EncodedInfo AudioEncoderOpusImpl::EncodeImpl(
       });
   input_buffer_.clear();
 
-  bool dtx_frame = (info.encoded_bytes <= 2);
-
   // Will use new packet size for next encoding.
   config_.frame_size_ms = next_frame_length_ms_;
 
@@ -625,15 +623,9 @@ AudioEncoder::EncodedInfo AudioEncoderOpusImpl::EncodeImpl(
   info.encoded_timestamp = first_timestamp_in_buffer_;
   info.payload_type = payload_type_;
   info.send_even_if_empty = true;  // Allows Opus to send empty packets.
-  // After 20 DTX frames (MAX_CONSECUTIVE_DTX) Opus will send a frame
+  info.speech = WebRtcOpus_GetInDtx(inst_) == 0;
-  // coding the background noise. Avoid flagging this frame as speech
-  // (even though there is a probability of the frame being speech).
-  info.speech = !dtx_frame && (consecutive_dtx_frames_ != 20);
   info.encoder_type = CodecType::kOpus;
 
-  // Increase or reset DTX counter.
-  consecutive_dtx_frames_ = (dtx_frame) ? (consecutive_dtx_frames_ + 1) : (0);
-
   return info;
 }
 

modules/audio_coding/codecs/opus/audio_encoder_opus.h

@@ -179,7 +179,6 @@ class AudioEncoderOpusImpl final : public AudioEncoder {
   std::optional<size_t> overhead_bytes_per_packet_;
   const std::unique_ptr<SmoothingFilter> bitrate_smoother_;
   std::optional<int64_t> bitrate_smoother_last_update_time_;
-  int consecutive_dtx_frames_;
 
   friend struct AudioEncoderOpus;
 };

modules/audio_coding/codecs/opus/opus_interface.cc

@@ -310,6 +310,16 @@ int16_t WebRtcOpus_GetUseDtx(OpusEncInst* inst) {
   return -1;
 }
 
+int16_t WebRtcOpus_GetInDtx(OpusEncInst* inst) {
+  if (inst) {
+    opus_int32 in_dtx;
+    if (ENCODER_CTL(inst, OPUS_GET_IN_DTX(&in_dtx)) == 0) {
+      return in_dtx;
+    }
+  }
+  return -1;
+}
+
 int16_t WebRtcOpus_EnableCbr(OpusEncInst* inst) {
   if (inst) {
     return ENCODER_CTL(inst, OPUS_SET_VBR(0));

modules/audio_coding/codecs/opus/opus_interface.h

@@ -245,6 +245,20 @@ int16_t WebRtcOpus_DisableDtx(OpusEncInst* inst);
  */
 int16_t WebRtcOpus_GetUseDtx(OpusEncInst* inst);
 
+/****************************************************************************
+ * WebRtcOpus_GetUseDtx()
+ *
+ * This function gets if the encoder is in DTX.
+ *
+ * Input:
+ *      - inst               : Encoder context
+ *
+ * Return value              :  0 - Encoder is not DTX.
+ *                              1 - Encoder is in DTX.
+ *                             -1 - Error.
+ */
+int16_t WebRtcOpus_GetInDtx(OpusEncInst* inst);
+
 /****************************************************************************
  * WebRtcOpus_EnableCbr()
  *