Page History: SIP audio settings
Compare Page Revisions
Page Revision: Thursday, 14 June 2012 10:38
The jitter buffer for incoming RTP (audio data being received from the SIP provider) is set centrally on the server using the server.ini file;
[audio]
jitterMs=20
Fine tuning of the audio server can be made locally on each audio server using the audioserver.ini file;
[audio]
WaveInPrefetch=20
WaveInBufferSize=50
WaveInMaxBuffers=2
WaveOutPrequeue=20
WaveOutMaxBufferSize=100
WaveInPrefetch
Controls the number of milliseconds of data that should be received from the audio input before allowing that data to be used. This prevents under-runs of the audio input.
This value does not normally need changing.
WaveInBufferSize
WAVE driver
Controls the buffer size (milliseconds) used to retrieve audio from the sound card. Larger buffers will cause more latency whilst they are filled before being handed to the audio server. Most consumer sound cards cannot cope with a buffer smaller than 50ms.
ASIO driver
WaveInBufferSize has no effect when using ASIO. The ASIO buffer size is set in the soundcards ASIO drivers. The value should be set to 960 which equates to 20ms or the highest value less than 960 if 960 cannot be selected. Usually this means 768.
WaveInMaxBuffers
The maximum buffers that can be stored before discarding wave in data.
This is a multiplier of the WaveInBufferSize. E.g. a setting of WaveInBufferSize=50 and WaveInMaxBuffers=2 would be a maximum wave in buffer size of 100ms. This must be at least twice that of any RTP packets that are being processed.
WaveOutPrequeue
Controls the number of milliseconds of audio that should be queued to the audio output before allowing it to play. This prevents under-runs of the audio output.
WaveOutMaxBufferSize
Controls the maximum number of milliseconds of audio that should be queued for playback. This prevents the output latency growing excessively over time if the clock of the soundcard varies slightly to that of the PC.
RTPInPrefetch
This is the amount of time the Audio Server will prequeue before starting to send audio to the soundcard. This lets the Audio Server build a buffer before starting audio. Default is 20ms.***Please also see jitterMS.
jitterMs
JitterMS is the total size of the jitter buffer. Bionics recommend that this should be twice the size of the RTPInPrefetch value. Default is 20ms.