Kodi Community Forum
AudioEngine branch - DO NOT REQUEST BINARY BUILDS - Printable Version

+- 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: AudioEngine branch - DO NOT REQUEST BINARY BUILDS (/showthread.php?tid=78289)



- liquidskin76 - 2011-04-05

jd2157 Wrote:Anyone know if there's anything else left on the todo list (other than core audio) before the merging will start? Just trying to (politely Smile ) get an idea if we're looking at weeks or months in terms of ETA.

HD audio bitstreaming?!!


- SpectreX - 2011-04-05

Hmm, i think topfs2 meant that it will be merged once it`s capable of bitstreaming the core audio (DTS/DD) like the present implementation, and HD audio can be worked on afterwards, until the final Eden release. Or at least that`s what i understood from his post.


- topfs2 - 2011-04-05

SpectreX Wrote:Hmm, i think topfs2 meant that it will be merged once it`s capable of bitstreaming the core audio (DTS/DD) like the present implementation, and HD audio can be worked on afterwards, until the final Eden release. Or at least that`s what i understood from his post.

I meant when core audio (alsa or directsound for osx) then its ready for merge, or atleast review.


- Anssi - 2011-04-05

Issues I'd personally like to see tackled (either by me or someone with more time available), here goes:
  • Regression of http://trac.xbmc.org/ticket/9248 (probably want to make the Sink have the channel layout from the AEStream if available; things get more complicated when upmix support is added as then we wouldn't necessarily want this always-on anymore)
  • Regression of http://trac.xbmc.org/ticket/10297 (FMT_RAW should probably be splitted to LE and BE varieties so that AE can handle byteswap as with other formats (or alternatively use AE_FMT_S16LE/AE_FMT_S16BE for raw as well and instead use m_encodedFormat != AE_ENCFMT_PCM to flag encoded audio), and we can query CoreAudio for the endianness it wants)
  • Regression of "Changed SL & SR channels to be forced to BL & BR when the source does not have BL & BR channels and the speaker layout does not have SL & SR" (i.e. this doesn't happen anymore; originally added to master in fcb1658d6) (should be a simple matter of mangling the channel layout a bit)
Now, I'm not against merging before these points are taken care of, though, just thought I'd write these up in case someone is interested fix them before I'll have time to take a closer look Smile


- davilla - 2011-04-05

AE's usage of LibTag needs addressing.


- topfs2 - 2011-04-05

davilla Wrote:AE's usage of LibTag needs addressing.

wow it uses that?! Don't really understand why it would need that?


- jmarshall - 2011-04-05

It doesn't. I thought it was in a separate branch. If not, it should be.


- SpectreX - 2011-04-05

jmarshall Wrote:It doesn't. I thought it was in a separate branch. If not, it should be.

Are you talking about not merging AE with the main branch?


- s3verian - 2011-04-06

topfs2 Wrote:I meant when core audio (alsa or directsound for osx) then its ready for merge, or atleast review.

Are there any open issues with the Windows implementation? As I understand it, on Windows, XBMC will be using FFDShow like MPC ala: http://www.mediasmartserver.net/2010/02/02/guide-setting-up-bitstreaming-with-your-windows-7-htpc-part-i/

Other than that, AE just needs HD audio codec detection and GUI options for bitstreaming (again this is just my understanding from reading the AE threads, I'm not a dev).

I'm currently running on the Windows nightly builds, and am very much looking forward to being able to test these features out Smile


- liquidskin76 - 2011-04-06

s3verian Wrote:Are there any open issues with the Windows implementation? As I understand it, on Windows, XBMC will be using FFDShow like MPC ala: http://www.mediasmartserver.net/2010/02/02/guide-setting-up-bitstreaming-with-your-windows-7-htpc-part-i/

Other than that, AE just needs HD audio codec detection and GUI options for bitstreaming (again this is just my understanding from reading the AE threads, I'm not a dev).

I'm currently running on the Windows nightly builds, and am very much looking forward to being able to test these features out Smile

ffdshow will be used on the dscodec development. It's a separate project.

AE will include HD audio bitstreaming however it will be native to XBMC.

Cheers


- sebak - 2011-04-07

SpectreX Wrote:Are you talking about not merging AE with the main branch?

I think he means that the integration of TagLib should be done in a seperate git branch than AE.
It probably is, but the dev of AE talked about it a couple of times in this thread.


- gnif - 2011-04-12

Ok, I am sorry I have not updated this thread in ages, RL and PC issues have prevented me from working on AE in quite a while...

A few things have been fixed in the last few days, a crash bug when transcoding to AC3 has been corrected, and gimli has found and fixed a bug in the DTS parsing, so DTS should now be working (not DTS-MA).

I have removed the setting apply delay again, this was more of a hack to try to work around a race condition I could not find, this bug was corrected about 20 minutes ago Smile.

Re TagLib.. there is a seperate branch for it in my repository, but I never finished it due to the missing ability to support a VFS, apparently the next version of TagLib will have an abstraction API, so it is delayed by that.

@Anssi, I will look at those regressions.

gimli has been assisting in adding iOS and CoreAudio support, I believe he has CoreAudio working already, but there is some issues that need to be addressed for embedded platforms as the float performance on them sucks. We are investigating a compile time option to switch XBMC back into S16 mode, but this is quite a major task in itself, so it will delay things a little.

Anyway, I will try to continue to push out more updates and information into this thread Smile


- liquidskin76 - 2011-04-12

Thanks for the update gnif.


- crash123 - 2011-04-12

gnif Wrote:Re TagLib.. there is a seperate branch for it in my repository, but I never finished it due to the missing ability to support a VFS, apparently the next version of TagLib will have an abstraction API, so it is delayed by that.
http://developer.kde.org/~wheeler/taglib/api/

Added in TagLib 1.7.0?


- gnif - 2011-04-12

@re taglib, its low priority at this point, if 1.7 has the ability to abstract the FS, then for sure, I will implement it post AE.