Push silence when re-creating audio device

Should minimize re-slaving clock jitter.

2 files changed, 43 insertions, 0 deletions

diff --git a/src/AudioRenderer.cpp b/src/AudioRenderer.cpp
index b0bdd3f..ed3518f 100644
--- a/src/AudioRenderer.cpp
+++ b/src/AudioRenderer.cpp
@@ -238,6 +238,7 @@ namespace SaneAudioRenderer
                 m_sampleCorrection.NewDeviceBuffer();

                 InitializeProcessors();

                 m_startClockOffset = m_sampleCorrection.GetLastFrameEnd();

+                MinimizeReslavingJitter();

                 StartDevice();

             }

             else

@@ -485,7 +486,10 @@ namespace SaneAudioRenderer
             m_startClockOffset = m_sampleCorrection.GetLastFrameEnd();

 

             if (m_state == State_Running)

+            {

+                MinimizeReslavingJitter();

                 StartDevice();

+            }

         }

     }

 

@@ -501,6 +505,43 @@ namespace SaneAudioRenderer
         }

     }

 

+    void AudioRenderer::MinimizeReslavingJitter()

+    {

+        assert(m_device);

+        assert(m_state == State_Running);

+        assert(m_device->GetEnd() == 0);

+

+        if (m_device->IsBitstream())

+            return;

+

+        // Try to keep inevitable clock jerking to a minimum after re-slaving.

+

+        REFERENCE_TIME silence = m_startClockOffset - (m_myClock.GetPrivateTime() - m_startTime);

+

+        if (!m_device->IsExclusive())

+            silence -= m_device->GetStreamLatency();

+

+        if (silence > 0)

+        {

+            silence = std::min(silence, llMulDiv(m_device->GetBufferDuration(), OneSecond, 1000, 0));

+

+            uint32_t rate = m_device->GetWaveFormat()->nSamplesPerSec;

+            DspChunk chunk(m_device->GetDspFormat(), m_device->GetWaveFormat()->nChannels,

+                           (size_t)llMulDiv(silence, rate, OneSecond, 0), rate);

+

+            if (!chunk.IsEmpty())

+            {

+                m_startClockOffset -= llMulDiv(chunk.GetFrameCount(), OneSecond, rate, 0);

+

+                DebugOut("AudioRenderer insert", chunk.GetFrameCount() * 1000. / rate,

+                         "ms of silence to minimize re-slaving jitter");

+

+                ZeroMemory(chunk.GetData(), chunk.GetSize());

+                PushToDevice(chunk, nullptr);

+            }

+        }

+    }

+

     void AudioRenderer::ApplyClockCorrection()

     {

         CAutoLock objectLock(this);

diff --git a/src/AudioRenderer.h b/src/AudioRenderer.h
index a09de1b..d90a4f7 100644
--- a/src/AudioRenderer.h
+++ b/src/AudioRenderer.h
@@ -65,6 +65,8 @@ namespace SaneAudioRenderer
         void CreateDevice();

         void ClearDevice();

 

+        void MinimizeReslavingJitter();

+

         void ApplyClockCorrection();

 

         void ApplyRateCorrection(DspChunk& chunk);