MediaPortal Audio renderer - better video playback quality (10 Viewers)

[red5goahead]

Tourettes.

Interesting behaviour with the 25 fps material on 75 HZ refresh rate.
the video hastes too. so the audio delay is not related to speedup to PAL issue.

 

[tourettes]

Tourettes.

Interesting behaviour with the 25 fps material on 75 HZ refresh rate.
the video hastes too. so the audio delay is not related to speedup to PAL issue.

Does the issue happen as well if you try some other audio renderer with the same dshowhelper.dll revision?

 

[tourettes]

I'm trying to play 25 fps PAL live TV with this patch. Setting the monitor to 60 Hz is OK but when trying 50 Hz the audio is all messed up. It's looping every second about three times. When watching tennis is sounds like he hits the ball three times.

Does that happen as well with EnableTimestretching disabled? or with ForceDirectSound enabled?

I think I need to write some troubleshooting quideline so I don't have to ask same questions on every post / test result

 

[tourettes]

Re: AW: MediaPortal Audio renderer - better video playback quality

1080p24 ts file on 50Hz display:

How was the actual playback quality? Stat OSD is broken, but seems that it has something to do with IONs since few have already reported such (no clue what could cause it).

 

[vuego]

Does that happen as well with EnableTimestretching disabled? or with ForceDirectSound enabled?

I think I need to write some troubleshooting quideline so I don't have to ask same questions on every post / test result

EnableTimestretching on or off = same problem
ForceDirectSound on = normal audio

 

[OnkelChris]

AW: Re: AW: MediaPortal Audio renderer - better video playback quality

How was the actual playback quality? Stat OSD is broken, but seems that it has something to do with IONs since few have already reported such (no clue what could cause it).

playback was fine... nothing to mention. everything was smooth and in sync...

the stats are not broken, not for me. i did some changes to the code, so the ION isn't dropping frames with the stats on. (just reduced the stats and disabled the refresh rate estimation, which didn't work well for me)

 

[red5goahead]

Tourettes.

Interesting behaviour with the 25 fps material on 75 HZ refresh rate.
the video hastes too. so the audio delay is not related to speedup to PAL issue.

Does the issue happen as well if you try some other audio renderer with the same dshowhelper.dll revision?

If I change the new audio renderer with reclock the play became fine (dshowhelper.dl don't change)

I attach an mp3 file.

I start two times about 1 min and 30 seconds using new audio renderer and reclock audio rendered. Using Audacity audio tools I've recorded the wave channel both times. so I put the 2 stereo track into the same project and i had synch them. the mp3 in synched at the beginning but it have a progressive desynch. On left channel there is the reclock track , on the right channel the new audio renderer track that end about 1 second / 1 second and half after the other one.

 

[tourettes]

Does that happen as well with EnableTimestretching disabled? or with ForceDirectSound enabled?

I think I need to write some troubleshooting quideline so I don't have to ask same questions on every post / test result

EnableTimestretching on or off = same problem
ForceDirectSound on = normal audio

Ok, then it is some compatibility issue with the WASAPI drivers and audio renderer (which will be really hard to debug since I dont have access to more than one audio chip...). What audio HW you are using?

 

[tourettes]

Re: AW: Re: AW: MediaPortal Audio renderer - better video playback quality

How was the actual playback quality? Stat OSD is broken, but seems that it has something to do with IONs since few have already reported such (no clue what could cause it).

playback was fine... nothing to mention. everything was smooth and in sync...

the stats are not broken, not for me. i did some changes to the code, so the ION isn't dropping frames with the stats on. (just reduced the stats and disabled the refresh rate estimation, which didn't work well for me)

Oh, now I see... please don't post test results with modified binaries unless you are explicitly stating in the posting exactly what has been modified. It is already too hard for me to guess from peoples error descriptions what might be happening (describing audio / video issues with a written language is not an easy task. Let alone understanding it).

So, for you it is working nicely currently? Could you share your codec and HW setup? We need to try to collect some information on the setups that are currently working and the ones that are failing.

 

[tourettes]

Tourettes.

Interesting behaviour with the 25 fps material on 75 HZ refresh rate.
the video hastes too. so the audio delay is not related to speedup to PAL issue.

Does the issue happen as well if you try some other audio renderer with the same dshowhelper.dll revision?

If I change the new audio renderer with reclock the play became fine (dshowhelper.dl don't change)

I attach an mp3 file.

I start two times about 1 min and 30 seconds using new audio renderer and reclock audio rendered. Using Audacity audio tools I've recorded the wave channel both times. so I put the 2 stereo track into the same project and i had synch them. the mp3 in synched at the beginning but it have a progressive desynch. On left channel there is the reclock track , on the right channel the new audio renderer track that end about 1 second / 1 second and half after the other one.

Could you check as well with the default directsound (or similar) audio renderer and compare the duration to reclock and mp audio renderer results? Since there is huge amount of audio sync adjustments done (4700+ frames drawn vs 137 sync adjustments) it is highly likely that Reclock is resampling the audio stream as well.

About the issue, is the audio drifting just a bit (lip sync getting off)? This might be aused by the fact that currently the reference clock is generated on the system clock, not on the audio clock and if those two are drifting... I haven't yet managed to implement the reference clock that would be based on the WASAPI playback buffer position.

One possibility would be the 1.0000000238 as bias value. It should be 1.0 since 3x 25.0 is 75.0. So, either the source has a bit wrong FPS value or some float calculation has issues with the accurace (althou it should be affecting the 50Hz as well).

Could you 1) provide the EVR and audio renderer logs for 75Hz and 50Hz cases 2) test how it behaves on 50Hz