2023-02-08, 03:20
Fixes have been implemented into the v.21 betas, and the nightly builds. Install the 32-bit build as Fire TV doesn't support 64-bit.
2023-02-08, 07:51
Amazon fixing their firmware bug? :-( So in 60 hz all good, but in 24p all off. Awesome. Sounds even worse than their Dolby processing delay which only was visible from the gettimestamp API... but that's already in.
Please start with a Debug Log. One hint: Kodi has advanced settings to only set a delay for 24p mode, that might workaround their issue, but only if this out of sync is constant. How much is it? The rest I most likely see in your debug log.
In any case try the latest aerework build first. It won't change much for now but this is where development continues.
Please start with a Debug Log. One hint: Kodi has advanced settings to only set a delay for 24p mode, that might workaround their issue, but only if this out of sync is constant. How much is it? The rest I most likely see in your debug log.
In any case try the latest aerework build first. It won't change much for now but this is where development continues.
2023-02-08, 10:42
Is it really true that (after an upgrade to the newest firmware) the Fire TV Cube 3 can passthrough all Audio formats with Kodi?
I doubt a little bit because all other Fire TV devices are not able to do it !
I
I doubt a little bit because all other Fire TV devices are not able to do it !
I
2023-02-08, 18:33
2023-02-08, 18:34
2023-02-08, 20:59
See what the OP tells - for now it looks like - yet another thing - that needs repairing :-)
2023-02-08, 21:26
its case by case based on file on gen3
right now, audio offset +0.125 is working for an 8 channel truehd atmos stream passthrough
the next movie will need something else
the movie after that wont work at all
no i cannot provide debug logs because logd is somehow broken and stuck on info level in the gen3
right now, audio offset +0.125 is working for an 8 channel truehd atmos stream passthrough
the next movie will need something else
the movie after that wont work at all
no i cannot provide debug logs because logd is somehow broken and stuck on info level in the gen3
2023-02-08, 21:39
enabel component logging for audio - and provide this. That will tell us the measured delays.
Make sure to disable it again afterwards quite fast, cause this spams your disk
So: open one of the formats affected, let it run for 1 minute and post this log
Make sure to disable it again afterwards quite fast, cause this spams your disk
So: open one of the formats affected, let it run for 1 minute and post this log
2023-02-09, 01:52
@fritsch, thanks for your feedback. Here's a debug log w/Kodi IEC packer option, adjust display refresh enabled, synch playback disabled that exhibits the audio synch issue. Played two 24p clips, each for a minute, the first w/DTS:X & the second w/TrueHD ATMOS.
2023-02-09, 08:01
I see that you are running a build of someone. I have no idea what it includes and what it patches. So that we are safe in our assumptions, you could please install official Kodi nightly build v21 next to it and do the log again?
The last builds I have seen, that not only added features, but pulled in fixes, which were sadly technically absolutely wrong.
The nightly currently has a very good testing for passthrough delays, therefore I suggest doing this. Your maven build won't be affected.
The last builds I have seen, that not only added features, but pulled in fixes, which were sadly technically absolutely wrong.
The nightly currently has a very good testing for passthrough delays, therefore I suggest doing this. Your maven build won't be affected.
2023-02-09, 09:37
DTS-X:
Normal DTS should have the same issue and most likely also AC3. Not sure they ever tested IEC on their own machines :-(. We can see clearly here that the sink is not accepting / forwarding any data for as much as 5 seconds - all the packages land on the HAL / surfaceflinger ... no HEAD moves.
Second Example:
TrueHD:
TrueHD actually looks okay. So from the logs I would say your DTS-X sample (which is player as DTS-Core) is seconds off - TrueHD Example looks okay. No large errors, nothing.
So what to do:
0) Set Keep alive Audio including low noise volume to at least 1 minute. Else Android will constantly reopen the sink, up to three times before the movie is actually started.
1) Try to switch to 3840x2160p IN the menu already and disable AdjustRefreshrate. Wait 5 seconds and play your sample again. Better worse?
2) If 1) has success try to tune the Refreshrate Delay to as large as 6 seconds, please.
-> and please: next debuglogs with an official nightly build. There are currently so many "builds" around here, that I lost the overview and many things are "short hacked" for a short-term success, but we cannot use these for debugging ... I hope you understand that. If the author of those builds wants to do that on his codebasis -> fell free to reach to him. I am very happy for every upstream contribution.
Quote:2023-02-08 18:29:07.415 T:17525 info <general>: Framecounter: 2816 Time: 82603151838762 Current-Time: 82603307296428It is opened btw. as DTS-Core. But that says it all I think. After having added 5.2 seconds (!) of data - both of the Delay APIs of android report: 214 ms played ... so your audio should be seconds (!) off.
2023-02-08 18:29:07.415 T:17525 info <general>: Delay - Timestamp: 4988.374999999931 (ms) delta: 155.627916 (ms) playtime: 214.29166666666666 (ms) Duration: 5202.666666666598 ms
2023-02-08 18:29:07.415 T:17525 info <general>: Head-Position 9728 Timestamp Position 10286 Delay-Offset: 3.843071682022823e+17 ms
2023-02-08 18:29:07.415 T:17525 info <general>: HW-Delay (1): -11.62500000000044 ms
2023-02-08 18:29:07.415 T:17525 info <general>: Combined Delay: 4988.374999999931 ms
Normal DTS should have the same issue and most likely also AC3. Not sure they ever tested IEC on their own machines :-(. We can see clearly here that the sink is not accepting / forwarding any data for as much as 5 seconds - all the packages land on the HAL / surfaceflinger ... no HEAD moves.
Second Example:
Quote:2023-02-08 18:29:07.262 T:17525 info <general>: Framecounter: 2816 Time: 82603151838762 Current-Time: 82603154779178
2023-02-08 18:29:07.263 T:17525 info <general>: Delay - Timestamp: 4895.624999999938 (ms) delta: 3.050458 (ms) playtime: 61.70833333333333 (ms) Duration: 4957.333333333271 ms
2023-02-08 18:29:07.263 T:17525 info <general>: Head-Position 2816 Timestamp Position 2962 Delay-Offset: 3.843071682022823e+17 ms
2023-02-08 18:29:07.263 T:17525 info <general>: HW-Delay (1): -3.0416666666672754 ms
2023-02-08 18:29:07.263 T:17525 info <general>: Combined Delay: 4895.624999999938 ms
2023-02-08 18:29:07.263 T:17525 info <general>: Delay Current: 4895.368056 ms
TrueHD:
Quote:2023-02-08 18:31:46.587 T:17525 info <general>: Framecounter: 10041344 Time: 82761885745129 Current-Time: 82762479205671
2023-02-08 18:31:46.587 T:17525 info <general>: Delay - Timestamp: 153.88541666802524 (ms) delta: 593.565667 (ms) playtime: 26446.114583333332 (ms) Duration: 26600.000000001357 ms
2023-02-08 18:31:46.587 T:17525 info <general>: Head-Position 10146816 Timestamp Position 10155308 Delay-Offset: 9.607679205057054e+16 ms
2023-02-08 18:31:46.587 T:17525 info <general>: HW-Delay (1): -22.114583333333826 ms
2023-02-08 18:31:46.587 T:17525 info <general>: Combined Delay: 153.88541666802524 ms
2023-02-08 18:31:46.587 T:17525 info <general>: Delay Current: 151.362847 ms
TrueHD actually looks okay. So from the logs I would say your DTS-X sample (which is player as DTS-Core) is seconds off - TrueHD Example looks okay. No large errors, nothing.
So what to do:
0) Set Keep alive Audio including low noise volume to at least 1 minute. Else Android will constantly reopen the sink, up to three times before the movie is actually started.
1) Try to switch to 3840x2160p IN the menu already and disable AdjustRefreshrate. Wait 5 seconds and play your sample again. Better worse?
2) If 1) has success try to tune the Refreshrate Delay to as large as 6 seconds, please.
-> and please: next debuglogs with an official nightly build. There are currently so many "builds" around here, that I lost the overview and many things are "short hacked" for a short-term success, but we cannot use these for debugging ... I hope you understand that. If the author of those builds wants to do that on his codebasis -> fell free to reach to him. I am very happy for every upstream contribution.
