Smooth voice power level reporting.

3 files changed, 58 insertions, 20 deletions

diff --git a/indra/llwebrtc/llwebrtc.cpp b/indra/llwebrtc/llwebrtc.cpp
index a3dadd696c..7b5ca7fb16 100644
--- a/indra/llwebrtc/llwebrtc.cpp
+++ b/indra/llwebrtc/llwebrtc.cpp
@@ -39,6 +39,7 @@ namespace llwebrtc
 
 const float VOLUME_SCALE_WEBRTC = 3.0f;
 
+LLAudioDeviceObserver::LLAudioDeviceObserver() : mMicrophoneEnergy(0.0), mSumVector {0} {}
 
 double LLAudioDeviceObserver::getMicrophoneEnergy() { return mMicrophoneEnergy; }
 
@@ -48,14 +49,26 @@ void LLAudioDeviceObserver::OnCaptureData(const void    *audio_samples,
                                          const size_t   num_channels,
                                          const uint32_t samples_per_sec)
 {
-    double       energy  = 0;
+    float       energy  = 0;
     const short *samples = (const short *) audio_samples;
     for (size_t index = 0; index < num_samples * num_channels; index++)
     {
-        double sample = (static_cast<double>(samples[index]) / (double) 32768);
+        float sample = (static_cast<float>(samples[index]) / (float) 32768);
         energy += sample * sample;
     }
-    mMicrophoneEnergy = std::sqrt(energy);
+
+    // smooth it.
+    size_t buffer_size = sizeof(mSumVector) / sizeof(mSumVector[0]);
+    float totalSum = 0; 
+    int i;
+    for (i = 0; i < (buffer_size - 1); i++)
+    {
+        mSumVector[i] = mSumVector[i + 1];
+        totalSum += mSumVector[i];
+    }
+    mSumVector[i] = energy;
+    totalSum += energy;
+    mMicrophoneEnergy = std::sqrt(totalSum / (num_samples * buffer_size));
 }
 
 void LLAudioDeviceObserver::OnRenderData(const void    *audio_samples,
@@ -479,7 +492,7 @@ void LLWebRTCImpl::setSpeakerVolume(float volume)
 
 double LLWebRTCImpl::getAudioLevel()
 {
-    return mAudioDeviceObserver->getMicrophoneEnergy();
+    return 20*mAudioDeviceObserver->getMicrophoneEnergy();
 }
 
 //
diff --git a/indra/llwebrtc/llwebrtc_impl.h b/indra/llwebrtc/llwebrtc_impl.h
index da829e52af..70586f9013 100644
--- a/indra/llwebrtc/llwebrtc_impl.h
+++ b/indra/llwebrtc/llwebrtc_impl.h
@@ -66,6 +66,8 @@ namespace llwebrtc
 class LLAudioDeviceObserver : public webrtc::AudioDeviceDataObserver
 {
   public:
+    LLAudioDeviceObserver();
+
     double getMicrophoneEnergy();
 
     void OnCaptureData(const void    *audio_samples,
@@ -81,7 +83,8 @@ class LLAudioDeviceObserver : public webrtc::AudioDeviceDataObserver
                       const uint32_t samples_per_sec) override;
 
   protected:
-    double mMicrophoneEnergy;
+    float mSumVector[30];  // 300 ms of smoothing
+    float mMicrophoneEnergy;
 };
 
 class LLWebRTCImpl : public LLWebRTCDeviceInterface,
diff --git a/indra/newview/llvoicewebrtc.cpp b/indra/newview/llvoicewebrtc.cpp
index 3073a58183..c22246d841 100644
--- a/indra/newview/llvoicewebrtc.cpp
+++ b/indra/newview/llvoicewebrtc.cpp
@@ -87,6 +87,7 @@ namespace {
     const F32 VOLUME_SCALE_WEBRTC = 0.01f;
 
     const F32 SPEAKING_TIMEOUT = 1.f;
+    const F32 SPEAKING_AUDIO_LEVEL = 0.05;
 
     static const std::string VOICE_SERVER_TYPE = "WebRTC";
 
@@ -2366,16 +2367,28 @@ void LLWebRTCVoiceClient::sendPositionAndVolumeUpdate(void)
 
     if (mWebRTCDataInterface && mWebRTCAudioInterface)
     {
-        uint32_t audio_level = (uint32_t) ((F32) mWebRTCDeviceInterface->getAudioLevel() * 256);
-        if (audio_level != mAudioLevel)
+        F32 audio_level = 0.0;
+        
+		if (!mMuteMic)
+        {
+            audio_level = (F32) mWebRTCDeviceInterface->getAudioLevel();
+        }
+        uint32_t uint_audio_level = (uint32_t) (audio_level * 256);
+        if (uint_audio_level != mAudioLevel)
         {
             Json::FastWriter writer;
-            Json::Value      root;
-            root["p"]             = (uint32_t) ((F32) mWebRTCDeviceInterface->getAudioLevel() * 256);
+            Json::Value  root;
+            root["p"]             = uint_audio_level;
             std::string json_data = writer.write(root);
 
             mWebRTCDataInterface->sendData(json_data, false);
-            mAudioLevel = audio_level;
+            mAudioLevel                       = uint_audio_level;
+            participantStatePtr_t participant = findParticipantByID(gAgentID);
+            if (participant) 
+			{
+                participant->mPower = audio_level;
+                participant->mIsSpeaking = participant->mPower > SPEAKING_AUDIO_LEVEL;  
+			}
         }
     }
 	
@@ -2677,10 +2690,12 @@ void LLWebRTCVoiceClient::OnDataReceived(const std::string& data, bool binary)
                 {
                     removeParticipantByID(agent_id);
                 }
-                participant->mPower      = (F32) (voice_data[participant_id].get("p", Json::Value(participant->mPower)).asInt()) / 256;
+                F32 energyRMS = (F32) (voice_data[participant_id].get("p", Json::Value(participant->mPower)).asInt()) / 256;
+				// convert to decibles
+                participant->mPower = energyRMS;
                 /* WebRTC appears to have deprecated VAD, but it's still in the Audio Processing Module so maybe we
 				   can use it at some point when we actually process frames. */
-                participant->mIsSpeaking = participant->mPower > 0.05;  
+                participant->mIsSpeaking = participant->mPower > SPEAKING_AUDIO_LEVEL;  
 			}
         }
     }
@@ -4446,10 +4461,16 @@ void LLWebRTCVoiceClient::leaveChannel(void)
 
 void LLWebRTCVoiceClient::setMuteMic(bool muted)
 {
+    participantStatePtr_t participant = findParticipantByID(gAgentID);
+    if (participant)
+	{
+        participant->mPower = 0.0;
+	}
     if (mWebRTCAudioInterface)
 	{
         mWebRTCAudioInterface->setMute(muted);
 	}
+    mMuteMic = muted;
 }
 
 void LLWebRTCVoiceClient::setVoiceEnabled(bool enabled)
@@ -4625,16 +4646,17 @@ BOOL LLWebRTCVoiceClient::getIsModeratorMuted(const LLUUID& id)
 	return result;
 }
 
-F32 LLWebRTCVoiceClient::getCurrentPower(const LLUUID& id)
-{		
-	F32 result = 0;
-    participantStatePtr_t participant(findParticipantByID(id));
-	if(participant)
+F32 LLWebRTCVoiceClient::getCurrentPower(const LLUUID &id)
+{
+    F32 result = 0;
+    if (!mMuteMic)
 	{
-        LL_WARNS("Voice") << "Power:" << participant->mPower << LL_ENDL;
-		result = participant->mPower*4;
+		participantStatePtr_t participant(findParticipantByID(id));
+		if (participant)
+		{
+			result = participant->mPower * 4;
+		}
 	}
-	
 	return result;
 }