[mkortstiege]

Yea, it's all about the back-end signaling new content and triggering the scan. Should have mentioned it earlier to avoid confusion.

[MrNice]

@FernetMenta

The future is near. First iteration of audio DSP has been submitted for review:
https://github.com/xbmc/xbmc/pull/4402

Within this context we think about how to manipulate the downmix matrix too.

Can you explain what MilhouseVH said here

This is not what I'd like and I am waiting FernetMenta add-on:

We intend let users configure the downmix matrix in a future version of XBMC.

I'd like to include that patch once it builds (and there are no other audio problems) however PR4402 is stale and unusable right now, and showing no signs of being updated (rebased).

Is the DSP development still on the way?

[Milhouse]

@MrNice: PR4402 no longer applies cleanly on tip of master. The PR is based on master from a month ago but master has already moved on with numerous other changes that have now invalidated many of the patches in the PR, meaning the PR has become "stale". The PR needs to be rebased so that it can apply cleanly once again on top of master, until then it's not possible to build XBMC with PR4402 its current state. I'm sure the rebasing will happen eventually. Of course I've no idea at this stage if these DSP changes will conflict with newclock3 changes, that's also quite probable.

[popcornmix]

@MrNice: PR4402 no longer applies cleanly on tip of master. The PR is based on master from a month ago but master has already moved on with numerous other changes that have now invalidated many of the patches in the PR, meaning the PR has become "stale". The PR needs to be rebased so that it can apply cleanly once again on top of master, until then it's not possible to build XBMC with PR4402 its current state. I'm sure the rebasing will happen eventually. Of course I've no idea at this stage if these DSP changes will conflict with newclock3 changes, that's also quite probable.

I'm not sure PR4402 actually does what was initially desired (custom setting of the mixing matrix). It was just indicated as the way in which this may be handled in the future.
I also imagine this may have a heavy CPU hit. While it could be possible to accelerate this in the future, I don't think this will use the current accelerated resampling/mixing code.

[Milhouse]

I'm not sure PR4402 actually does what was initially desired (custom setting of the mixing matrix). It was just indicated as the way in which this may be handled in the future.
I also imagine this may have a heavy CPU hit. While it could be possible to accelerate this in the future, I don't think this will use the current accelerated resampling/mixing code.

Fair enough, my plan had been to create a special "experimental" test build for the DSP commit as it looks like a fairly large and involved change and probably not appropriate for the "regular" test builds (at least not initially) but for all the reasons above I'll give it a miss for the time being (sorry MrNice).

Presumably if this does have a heavy CPU hit, if/when it lands on master it will need a mechanism to allow it to be disabled for lower powered devices?

[popcornmix]

Presumably if this does have a heavy CPU hit, if/when it lands on master it will need a mechanism to allow it to be disabled for lower powered devices?

Yes, looks like it is optional, and can be enabled/disabled in settings.

[MrNice]

No luck for me.
Could you advice me:
I use so far Openelec in my RPi with passthrough to only listen music. I have stereo and MCH (5.0 and 5.1) and my hardware is 5.0.
I have different level of quality from 44.1K to PCM, DTS and DD HD.
I want gapless play as well.
I see Openelec that doesn't fill the requirement.
What is the best software (XBMC, Raspbmc, other?) to use to get the full corresponding quality and which player?

Anyway many thanks for you great work.

[MrNice]

Following me above post, I was thinking:

TrueHD passthrough requires running the audio channel at 8 channel @ 192kHz which isn't supported by the hardware, so it is not possible.

But I think the Pi decode TrueHD is a very good job! I think we cannot hear a differennt(PCM Multichannel 5.1 vs. HD-Sound passthrough)

So I could use RPi decoding, no problem. Later I'll buy a fan-less computer.

popcornmix told me PAPlayer is the best choice in Openelec.
In previous build PCM mapping was good (in my opinion) LFE to FL/FR and Channel 7 and 8 to FL/RL and FR/RR.

1- There is almost nothing to do to get back to this setup.

2- The only issue I have is it doesn't play gapless, at least Multichannel, I didn't try PCM 16/44.1.

If these 2 above could be fixed, Openelec + PAPlayer could be perfect for my requirement.

Is it possible?

[popcornmix]

2- The only issue I have is it doesn't play gapless, at least Multichannel, I didn't try PCM 16/44.1.

omxplayer is based on dvdplayer which doesn't support gapless. It never will support gapless.

[MrNice]

2- The only issue I have is it doesn't play gapless, at least Multichannel, I didn't try PCM 16/44.1.

omxplayer is based on dvdplayer which doesn't support gapless. It never will support gapless.

Sorry I was not very clear, I will use only PAPlayer as you said it it better but PAPlayer doesn't play gapless today.

[popcornmix]

Sorry I was not very clear, I will use only PAPlayer as you said it it better but PAPlayer doesn't play gapless today.

It's not possible to change the number of speaker channels in a gapless way (you will get the receiver resync when channels change).
It may work if you change "output configuration" to fixed and enable stereo upmix.

[MrNice]

It's not possible to change the number of speaker channels in a gapless way (you will get the receiver resync when channels change).
It may work if you change "output configuration" to fixed and enable stereo upmix.

I don't understand what you mean.

I have multichannel music, mainly classical, live recorded or with parts (Symphony movements) sometime without gap or recorded live concert. There is no change in the speaker channels so this should play gapless.

Stereo upmix is to play a stereo file with (fake) multichannel, this is not what I do and want.

[popcornmix]

I don't understand what you mean.

I have multichannel music, mainly classical, live recorded or with parts (Symphony movements) sometime without gap or recorded live concert. There is no change in the speaker channels so this should play gapless.

Stereo upmix is to play a stereo file with (fake) multichannel, this is not what I do and want.

Can you test with those settings?

[MrNice]

I did a test with MilhouseVH build: #0411b
This is not the good thread but I answer your question.

For some reason, in the files I used to test, I can't reproduce the drop I had between gapless files, these files had this drop previously but I don't know which build.
However, with another symphony, at the 2nd movement (file), 10 seconds before the end, there are drops every 1-2 seconds, then at the end of the file, in full screen, visual time cursor stops, screen doesn't change (still 2nd movement) and music continue the next file (3rd movement) with drops.
Remote control is not active any more, mouse move and when I click to stop plying, mouse freezes, music still playing then without drops.
I have to disconnect the USB drive to stop playing.
I tried 3 times.
Same symphony between movement 1 and 2 plays gapless, change visual cursor to next file, OK.
Audio for movement 2 and 3:
Format : FLAC
Format/Info : Free Lossless Audio Codec
Duration : 7mn 18s
Bit rate mode : Variable
Bit rate : 6 528 Kbps and 6 650 Kbps and 6 381 Kbps
Channel count : 6 channels
Channel positions : Front: L C R, Back: L R, LFE
Sampling rate : 88.2 KHz
Bit depth : 24 bits

I didn't try stereo upmix as this is good with the usual file I test.

[MrNice]

I think I did a mismatch between OMXplayer and PAPlayer.
In Beta and MilhouseVH default is different.

So, PAPlayer has gapless working (but exception)
OMXplayer has no gapless but this is know and will remain lick that. No problem.

I tried again PAPlayer in v3.95.5 (this is not the good thread) and I have the same as describe above, only between movement 2 and 3.
I'll try with next version.