MediaPortal Audio renderer - better video playback quality (1 Viewer)

[GeYe]

AW: MediaPortal Audio renderer - better video playback quality

tourettes
Good work. I switched from PowerDVD Codec Mpeg2 TV to Microsoft Codec, because SAF 5 PowerDVD has some problems. I think the screenshot also shows this. PowerDVD Codec drops no frames, but is not smooth and has also stuttering while looking tv for long time.

Most frame drops come from switching channel, open menü or taking screenshots.

The logs are from normal TV viewing.

 

[red5goahead]

Could someone with a Xonar card try the version in this post and upload the resulting log file:

https://forum.team-mediaportal.com/...layback-quality-84441/index96.html#post674650

It may help fix the issues with problematic cards.

done.

11-11-2010 19:39:25.457 [ 11f8] Base Clock : 10474193 adjAdjustments : 10010 biasAdjustments : 448022 Audio Processed : 26996875 Audio Out : 28276874 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 0.990442 Drift : 387850.660428
11-11-2010 19:39:25.546 [ 11f8] Base Clock : 11408508 adjAdjustments : 12822 biasAdjustments : 488138 Audio Processed : 27870208 Audio Out : 29177916 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.022669 Drift : -552739.451472
11-11-2010 19:39:25.631 [ 11f8] Base Clock : 12249295 adjAdjustments : 15352 biasAdjustments : 524238 Audio Processed : 28713541 Audio Out : 30103958 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.012905 Drift : -236310.730783
11-11-2010 19:39:25.713 [ 11f8] Base Clock : 13025647 adjAdjustments : 17688 biasAdjustments : 557572 Audio Processed : 29576875 Audio Out : 31022916 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 0.964569 Drift : 1259501.375158
11-11-2010 19:39:25.803 [ 11f8] Base Clock : 13960691 adjAdjustments : 20501 biasAdjustments : 597719 Audio Processed : 30480208 Audio Out : 31908541 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.032885 Drift : -938787.517563
11-11-2010 19:39:25.892 [ 11f8] Base Clock : 14871265 adjAdjustments : 23241 biasAdjustments : 636816 Audio Processed : 31383541 Audio Out : 32794166 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.045052 Drift : -1427502.245266
11-11-2010 19:39:25.974 [ 11f8] Base Clock : 15693983 adjAdjustments : 25717 biasAdjustments : 672141 Audio Processed : 32236875 Audio Out : 33713749 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.003612 Drift : -46682.087109
11-11-2010 19:39:26.068 [ 11f8] Base Clock : 16600829 adjAdjustments : 28446 biasAdjustments : 711077 Audio Processed : 33100208 Audio Out : 34625624 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.012626 Drift : -352418.863519
11-11-2010 19:39:26.154 [ 11f8] Base Clock : 17479329 adjAdjustments : 31089 biasAdjustments : 748797 Audio Processed : 33963541 Audio Out : 35544583 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.009744 Drift : -245700.588292
11-11-2010 19:39:26.236 [ 11f8] Base Clock : 18309328 adjAdjustments : 33587 biasAdjustments : 784434 Audio Processed : 34826875 Audio Out : 36460416 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 0.962098 Drift : 1490162.572809
11-11-2010 19:39:26.332 [ 11f8] Base Clock : 19261107 adjAdjustments : 36451 biasAdjustments : 825300 Audio Processed : 35730208 Audio Out : 37346041 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.012790 Drift : -423698.736828
11-11-2010 19:39:26.413 [ 11f8] Base Clock : 20067158 adjAdjustments : 38876 biasAdjustments : 859909 Audio Processed : 36593541 Audio Out : 38265208 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 0.994068 Drift : 296029.219129
11-11-2010 19:39:26.504 [ 11f8] Base Clock : 20969451 adjAdjustments : 41591 biasAdjustments : 898650 Audio Processed : 37476875 Audio Out : 39165833 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 0.992884 Drift : 324537.185536
11-11-2010 19:39:26.585 [ 11f8] Base Clock : 21779695 adjAdjustments : 44029 biasAdjustments : 933439 Audio Processed : 38350208 Audio Out : 40068124 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 1.025378 Drift : -980958.851799
11-11-2010 19:39:26.679 [ 11f8] Base Clock : 22728420 adjAdjustments : 46884 biasAdjustments : 974174 Audio Processed : 39213541 Audio Out : 40986249 bias : 1.042808 adjustment : 1.003009 Drift Multiplier : 0.995794 Drift : 222294.332935

 

[davidf]

Could someone with a Xonar card try the version in this post and upload the resulting log file:

https://forum.team-mediaportal.com/...layback-quality-84441/index96.html#post674650

It may help fix the issues with problematic cards.

done.

How far out was the audio from the video (ish)?

I'm not actually very hopeful after reading the log.

 

[red5goahead]

Could someone with a Xonar card try the version in this post and upload the resulting log file:

https://forum.team-mediaportal.com/...layback-quality-84441/index96.html#post674650

It may help fix the issues with problematic cards.

done.

How far out was the audio from the video (ish)?

I'm not actually very hopeful after reading the log.

About 2 seconds after 26 minutes. (audio is delayed)

 

[davidf]

About 2 seconds after 26 minutes. (audio is delayed)

I've attached a picture of the drift reported by your Audio Card in ms - it should be a relatively straight line. I expected your card to show large variance at the begining and then settle down after that, as the misreporting seemed to correct itself at certain periods in the log Tourettes got you to produce earlier. It appears that this doesn't happen during playback and it continues to inaccurately estimate throughout.

The last 4 drift times in ms are:

65071.42329
-4452.583383
830.6678745
4372.093326

and they were all reported within 0.25 seconds

Unless I see something wrong in the current code then I couldn't get the drift corrected. The drift correction applied totalled 4 seconds so even disallowing drift would leave you 2 seconds out the other way.

I'll have a check at the code but I suspect that there is no way to fix the issue currently (unless the drivers change).:sorry:

Edit: Actually don't panic - I've spotted something...

 

[tourettes]

Edit: Actually don't panic - I've spotted something...

This post shows how the audio clock is lying to us.

https://forum.team-mediaportal.com/667280-post806.html

My assumption would be that the driver just incorrectly assumes the current playback position (on based some data which we dont see) and then corrects it the when it next time sees the real playback position.

 

[davidf]

My assumption would be that the driver just incorrectly assumes the current playback position (on based some data which we dont see) and then corrects it the when it next time sees the real playback position.

That was my assumption from the previous log data, that isn't the case (at least from a data point of view). I tracked it back last night and found the piece of code which prevents the drift code seing the correction made by the drivers when it sees the real playback position. I was going to remove it but I'm not sure how it will affect FF and RW so I'll leave it to you .

 

[gloomyandy]

Folks,
a quick question... What's the best way to insert a delay into the audio playback (to correct for my playback chain having audio ahead of video)? I can either use the ffdshow settings (what I've done in the past), or I can use the new delay setting... Which is the best way to do this?

Thanks

Andy

 

[tourettes]

Edit: Actually don't panic - I've spotted something...

Davidf's spotting results can be found in the attached binary. All Xonar users please test how it behaves. It has relaxed clock handling so it wont detect driver bugs for other HW (but backward jumping clock shouldn't even exist based on MSDN documentation )

So, Xonar owners please download and test (others don't need to bother with the download).

 

[tourettes]

a quick question... What's the best way to insert a delay into the audio playback (to correct for my playback chain having audio ahead of video)? I can either use the ffdshow settings (what I've done in the past), or I can use the new delay setting... Which is the best way to do this.

In theory there shouldn't be any difference if the delay is done on ffdshow or audio renderer side. But in practise it might cause dropped data when delayed too much on ffdshow side (depending how the actual delay is done on ffdshow side - if it just done by modifying the timestamps then it should be ok).

So, just test which one works the best (I would recommend audio renderer to be safe).