What phone app?
Does it remember the audio track on new playback?
Iāve been using the android app āAV Syncā. Bit painful to use on that particular sample though, need to hold my phone very close to the dots on the screen to get enough of a brightness change. Works much easier when using the Spears and Munsil clip with the white flashing bars.
Yes
I would not trust such app. I tried it and it shows what I see and hear, with TrueHD. Will check the other audio later.
I tested now all other audio layer and all are good.
When you say itās 7 frames off it would be a delay of 300ms. For this no app is needed to seeā¦
I canāt see such offset anymore in any case so you will need to make a video to show the offset.
Using last test build in clean setup environment without any addon for sure.
You also can check the current diff in player debug as live value.
1 Like
Iām not sure if I am understanding correctly, are you saying in your last test build the audio is in sync?
Yeah thatās what heās saying. @doppingkoala is suggesting the app he is using to measure is showing a 7 frame offset. @Portisch is stating heās not seeing that and is correctly advising that a 7 frame offset is so big it would be immediately apparent without measurement.
I didnāt have much luck with the app either, it seemed like I did not have enough granular control over the light and sound to really dial the app in.
I feel like Iām taking crazy pills with this thing. For the past month or so since I have started using this device and OS, some things are so far off itās obvious to me and my GF, but others not so much. Last night on the second test build she said the audio was in sync with her Love Island show, to me, it looked off but not too far off.
Is that the field that is āPlayer: a/v: Xā, where X is a number? In seconds?
Seconds, yes.
Like in my case itās 0.024 to 0.017.
1 Like
Are these changes likely to make the nightlies soon?
Here is a video showing a change in audio sync after a seek on the TrueHD track of the Spears and Munsil clip. Points that stick out to me and all shown in the video,
- Initially: Audio is late ~1 frame and the current diff in the player debug ranges from 0.097 to 0.189. Could not observe any change in audio sync either while playing the file or later examining the recording as the numbers in the player debug fluctuated.
- After a seek: Audio is ~5 frames late. Current diff in the player debug now is in the same range as you reported.
Done on your latest test build. Only addon is keymap editor I used to bring up the debug info.
edit: eac3 is definitely more consistent with itās audio sync now and I struggled to replicate an issue when recording. If I can replicate what I saw the other day while recording, I will upload that as well
You are kidding, isnāt it? 
You use a complete different sample video.
How I should debug what you might mixed up by converting to different media formats?
The original is Spears & Munsil UHD HDR Benchmark_t21.mkv.
You use a ***.m2ts.
Thanks again for loose free timeā¦
I recognize this filename to be one of mine, itās possible that @doppingkoala is using the same source in a different container because they extracted it themselves. Mine is extracted from the Spears and Munsil 2019 disk.
Okā¦
So clearly I am mistaken or missing something. Far as I knew, the file I used was in the original format, the .m2ts format is a format native to blurays, CE plays it, and other devices can skip in that file without changing 4 frames of sync. Given all that, I assumed it was a valid test file. So that I know what to avoid in the future, could you please let say what formats must be used ?
In any case, I do apologise for any time that you may have unnecessarily spent. Looking back through my posts, especially my ones interacting with you and providing logs, I think that video was the only one that used that sample though - so it hopefully wasnāt much time.
With some more testing on Audio_Video_Sync_23.976_TEST TrueHD DTS-HD DDP DD PCM.mkv and your latest build I am seeing a constant +1 offset for the audio with the TrueHD track on that sample. For all other audio formats in that sample however, it is mostly sitting at ~+5 for me. Obviously I canāt tell by myself if this difference is somehow an issue with my setup or caused by CE - so I am reporting it.
And thank you for the time you have spent on the audio sync problem, to me, your latest test build is a clear improvement over the currently nightly.
This is the reason. I only have this test file and all is in sync.
So when you have a offset somewhere else you need to provide the correct setup so I can reproduce it.
Here is the sample I used in the video. Could you reply after downloading so I can remove it.
Any thoughts on the difference between the TrueHD track and the others I am seeing? To me, it is suspicious that it is a 4 frame difference which exactly matches the display latency kodi assumes.
This is a dual layer sample?
So I guess kodi do match the two layers up in software and this is maybe just to much for SM1 SoC. I will test it next days.
1 Like
Far as I can tell its not. SDR video with multiple audio tracks is all according to mediainfo.
File not foundā¦
Unfortunately I am not in a position to help test out these development images at the moment, but if you are looking for other test media, please try 05. Audio-delay-test-v5_23.976.mkv from these test patterns (located in the Audio Delay and Bitstream folder). I find the issue to be easily reproducible with this file.