2011-04-15, 17:17
03.05.2011 UPDATE: The audio patch is now in the trunk. Please use Keith's official nightly builds.
Since I too like lukenukem experienced the same issue with a lot of "Discontinuity" errors in my log, I compiled my own build and then went back to the old atv2 branch on https://github.com/xbmc/atv2/commits/ and reverted back the IOSAudioRenderer.cpp until the sound started working. I pinpointed the problem to changes made to the AddPackets() method after the following commit: https://github.com/xbmc/atv2/commit/b0b7...0c39c344b0
If you add the check that makes sure there is free space in the audio buffer before accepting new packets, the problem goes away. In the code that can be done by adding the following code back to the AddPackets():
I hope this helps speed up the development. Big thanks to Scott, Zeljko and others who contributed to this great port of XBMC and to Keith for facilitating the "early" testing of the trunk branch.
lukenukem Wrote:I am running the latest nightly build.
The playback of certain video files is too fast. This makes watching the video unpleasant. As the video and audio do stay in sync, the audio is speeded up as well and gets distorted. (With noticable 'scratches' in the sound)
It's important to note that running the latest official build I could play the exact same files fine.
Since I too like lukenukem experienced the same issue with a lot of "Discontinuity" errors in my log, I compiled my own build and then went back to the old atv2 branch on https://github.com/xbmc/atv2/commits/ and reverted back the IOSAudioRenderer.cpp until the sound started working. I pinpointed the problem to changes made to the AddPackets() method after the following commit: https://github.com/xbmc/atv2/commit/b0b7...0c39c344b0
If you add the check that makes sure there is free space in the audio buffer before accepting new packets, the problem goes away. In the code that can be done by adding the following code back to the AddPackets():
Code:
if(len > free)
return 0;
I hope this helps speed up the development. Big thanks to Scott, Zeljko and others who contributed to this great port of XBMC and to Keith for facilitating the "early" testing of the trunk branch.