+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Development (https://forum.kodi.tv/forumdisplay.php?fid=32)
+--- Forum: Kodi Application (https://forum.kodi.tv/forumdisplay.php?fid=93)
+--- Thread: Testing audio engine ActiveAE (/showthread.php?tid=170338)
RE: Testing audio engine ActiveAE - fritsch - 2013-09-16
You should be on afb957
RE: Testing audio engine ActiveAE - fritsch - 2013-09-16
Do a git fetch and a reset --hard on that cuefile branch (or visit me on irc)
RE: Testing audio engine ActiveAE - FernetMenta - 2013-09-16
(2013-09-16, 21:33)franzem Wrote: sure: http://xbmclogs.com/show.php?id=59467
The System Information is: XBMC 13.0-Alpha Git.20130916-387c36c
where the 387c36c would indicate that I am using Fernet's latest version, as I say, I can see the behaviour is different from the log file.
I tested with the sample you provided and did not notice anything going wrong. Please provide the steps to reproduce the remaining issue.
RE: Testing audio engine ActiveAE - franzem - 2013-09-16
ok, that it very strange, let me eliminate all possible problems, and get back to you.
RE: Testing audio engine ActiveAE - herrnst - 2013-09-16
@FernetMenta
Re PR#3273, I've discovered this:
Generally, gapless playback with MP3/CUE works fine so far (no audible gap, info updating correctly etc.) This however should be reproducable:
After letting PAPlayer advance to the next track in the cuesheet, if you open the mediainfo popup (default: m-key) and seek around in the current track, PAPlayer always seeks to the "selected" offset _in the next track_, while pretending (infolabels suggest this) it still plays the current track (e.g. the track you were listening to before seeking).
Additionally, when the track was the last track in the cue file, you are suddenly in a sort of "hidden" track which really is the mp3 file as it wouldn't be played with the cue-information. In this state, when you seek again, you're seeking in the full file (imagine a 60minute file, when seeking to approx. 1/3, you're now playing from 00:20:00, while the track length still says e.g. 04:50).
I can reproduce this with any mp3/cue file and even flac/cue. This is with XBMC HEAD plus https://github.com/xbmc/xbmc/pull/3273.patch (patch starting with 387c36c), or your patched-up master branch.
Regards,
nst
RE: Testing audio engine ActiveAE - franzem - 2013-09-17
I am still struggling with a gap/jump transitions with my reference file, and I've noticed tiny jumps around the transitions in other files.
As mentioned, I can see from the log file that no files are being reloaded, so I'm pretty confident I have the correct source (either 387c36c or afb957, tried both).
Reproduce:
- fresh build: ccache, make clean, etc
- recompile, install, check System Info for version
- removed cue file to be sure, cleaned library, added files, updated library
- played track from scratch, small but perceptible jump track 1 to 2, large very obvious jump track 2 to 3
- usually played from Fullscreen Player, but same issue with Home screen
- disabled background Fanart
- Using Quartz skin, but same with Confluence
Anyway, looks like everybody else is getting the correct result, maybe I am doing something wrong, which is fine, you guys are doing great work on this ActiveAE and ultimately it's not about me getting some build right.
I think I'll let you get on with it and then maybe take another snapshot a few weeks from now, and I can give some more feedback then. If you would like me to do any additional testing or try some additional steps in the meantime please do not hesitate to ask.
RE: Testing audio engine ActiveAE - FernetMenta - 2013-09-17
I have updated the PR (squashed fix). Seeking did not work.
RE: Testing audio engine ActiveAE - herrnst - 2013-09-17
(2013-09-17, 08:42)FernetMenta Wrote: I have updated the PR (squashed fix). Seeking did not work.Will rebuild with the updated PR contents in the evening (after work etc
) and retry and report back. Cheers!
RE: Testing audio engine ActiveAE - herrnst - 2013-09-17
(2013-09-17, 09:15)herrnst Wrote:(2013-09-17, 08:42)FernetMenta Wrote: I have updated the PR (squashed fix). Seeking did not work.Will rebuild with the updated PR contents in the evening (after work etc
So, did a rebuild with 3273 (based around bedff7b) applied and tested, seeking now works inside the current track's boundaries on mp3+cue, flac+cue and dtswav+cue, so guess it's safe to assume the last round fixed it!
Thanks & cheers,
nst
RE: Testing audio engine ActiveAE - fritsch - 2013-09-17
I pushed it in the meantime.
RE: Testing audio engine ActiveAE - ursli - 2013-09-17
(2013-09-14, 11:56)fritsch Wrote: That is not an issue.
You just see Multi In on your AVR. ActiveAE opens the "user set channel mask" by default. You won't hear anything out of those speakers. That was design decision, if you read some pages back. (In short: stutter, when watching LiveTV and audio changed during commercials).
If you want the 2.0 behaviour, e.g. best sink for content, add the <audiophile>1</audiophile> option to your advancedsettings.xml
I hope you don't mind me having another question regarding the audiophile setting, it looks like mono audio (I've got some TV channels broadcasting in mono) is still being sent as lpcm even with audiophile set to 1, which in turn stops my receiver from upmixing it. Is that a restriction because of this special case or something going wrong on my end?
Log: http://www.xbmclogs.com/show.php?id=59812
RE: Testing audio engine ActiveAE - FernetMenta - 2013-09-18
should be fixed by https://github.com/xbmc/xbmc/pull/3289
RE: Testing audio engine ActiveAE - franzem - 2013-09-18
sorry to be bringing this up again, but I have just built the master of main xbmc trunk, 92169a, and I am still noticing jumping at the tracks transitions.
Either I am stupid at building xbmc from git (but I checked the source, it seems right correct, i can see the code changes; checked the git logs, I have the latest, did a ccache -cC , etc) or I am hearing things you guys aren't. In which case there might still be a problem.
Also, when I look at the log files I can see that the mp3 is being loaded at each transition, but I don't know if this is relevant to the stream being opened/closed or not.
I am attaching am log file, maybe that could shed some light: http://xbmclogs.com/show.php?id=60104
Or if you think of anything obvious in building the master or activating new behaviour, etc, I would be more than happy to do that. I suppose if I am doing something wrong, then I will evenvually get a working build. But if there is still something wrong with the cuefile handling, then now would be the time to address it...
this is my git log:
commit 92169a699b18a94b07d28e9d8a8aa5fd929c5f12
Merge: 0acee90 7093b1c
Author: Peter Frühberger <[email protected]>
Date: Tue Sep 17 11:22:28 2013 -0700
Merge pull request #3273 from FernetMenta/cuefile
paplayer fixes - all were tested by affected users in the testing thread
commit 0acee901787540127982d3ee7b2d0dc3f12abb31
Merge: 3757f41 72fdffd
Author: wsoltys <wiso@xbmc.org>
Date: Tue Sep 17 11:18:39 2013 -0700
Merge pull request #3285 from Skixbmc/WIN32Exception
[Win32] fixed: stack and dump writer couldn't handle smb paths
RE: Testing audio engine ActiveAE - FernetMenta - 2013-09-18
can you upload that file you tested with? what about the other file you provided, does this play ok?
RE: Testing audio engine ActiveAE - FernetMenta - 2013-09-18
ok, found it: you have added those to musicdb, in this case the path I currently compare does not match. will push a fix tomorrow.
btw: still want to have those Loco Dice tunes