W3C home > Mailing lists > Public > public-webrtc-logs@w3.org > September 2017

[webrtc-stats] jitterBufferDelay and concealed samples, DTX/CNG samples

From: henbos via GitHub <sysbot+gh@w3.org>
Date: Wed, 27 Sep 2017 07:57:32 +0000
To: public-webrtc-logs@w3.org
Message-ID: <issues.opened-260880604-1506499040-sysbot+gh@w3.org>
henbos has just created a new issue for https://github.com/w3c/webrtc-stats:

== jitterBufferDelay and concealed samples, DTX/CNG samples ==
jitterBufferDelay is total time each audio or video samples has spent in the jitter buffer before being decoded.

For video, average time spent in jitter buffer is jitterBufferDelay/framesDecoded. This makes sense.

For audio, it is said to be jitterBufferDelay/totalSamplesReceived. Possible problems:
- totalSamplesReceived include concealedSamples, which have spent no time in the jitter buffer. Should we instead divide by (totalSamplesReceived - concealedSamples)?
- We may have received samples that have not been decoded and ready to be played out, and these samples have not contributed to jitterBufferDelay yet. There is no "samplesDecoded", is totalSamplesReceived >= the value we should divide by?
- How are DTX/CNG samples handled? Do these contribute to jitterBufferDelay? Do these contribute to totalSamplesReceived?

Please view or discuss this issue at https://github.com/w3c/webrtc-stats/issues/246 using your GitHub account
Received on Wednesday, 27 September 2017 07:57:22 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 4 June 2019 15:32:45 UTC