Audio sync issue with capture device and SVP 'Auto' threads (Page 1 of 2)

Thanks, the audio delay in MPC-HC is having no effect though.

Other options on the audio switcher screen work though, such as Boost.

Yes, and I can't really tell if it makes a difference because it only allows +/- 250ms whereas my audio is a fair bit more out of sync.

Thanks, I added those values.

Unfortunately they have no affect whatsoever, for my setup, either + or -.

Thanks - well I tried the ffdshow audio processor, which support a Delay but ran into another problem - basically it seems to be set up ok in the filter graph, ie it says that it is taking the correct input and outputting it to the TV but no sound comes through.

The audio processor is definitely receiving the signal though because in the configuration panel if I tick 'Volume' and show levels I can see the L and R channels fluctuating.

For some reason though no sound actually goes to the audio device.

I will try the AC3Filter then.

Maybe the issue is because the capture is uncompressed audio or something.

Turns out that if I use a different USB capture device things change.

I was using one made by Inogeni and that was giving me the 750ms video latency with 25-to-50fps interpolation and hence it needed the same audio delay. Tried one made by Magewell and this reduces the latency to about 400ms, similar to 29-to-59fps interpolation.

That's strange because when I turn off SVP completely, both devices exhibit similar and very small latency.

Anyway, it got me thinking - I believe there is an AVISynth SetAudioDelay setting, why not add a setting for this for a given SVP profile? Even in my current scenario with the new device I get video latency of about 400ms which is enough to need an audio delay.

It would be much tidier to be able to tie this setting into the profile, rather than my evil hack to launch the separate Audio Repeater app.

Although, given my experience with setting the audio delay via MPC-HC's own settings I'm not convinced avisynth SetAudioDelay would work - perhaps one of the devs can point me in the direction of the AVISynth script that SVP creates so I can experiment with hardcoding some calls?

2) yes, I have reduced this from 3 to 1, as I found (by enabling AVISynth OSD) that it didn't need the extra two. It didn't make a significant difference though.

3) I will take a look. Is that a different buffer to the ffdshow ahead buffer?

But even so, is it not a potential flaw that depending on the hardware/threads etc the audio can be delayed due to the video latency?

I'm not really concerned about having a video latency of even up to 1s, if it makes things for SVP as good as possible, but I feel there needs to be a way of controlling the audio delay - but in such a way that it can be tuned on a per-profile basis, and also works for capture cards not just video files.

Would generate.js be the place to put in a call to SetAudioDelay too? Is that the place where it creates the AVISynth script that gets used?

I appreciate your help by the way!

oviano
Yes. generate.js is place where the Avisynth script is created. Chainik knows more about internal Avisynth buffer than me. He knows why 'do not try 3rd option'

In any way SVP needs video buffer to process frames and to produce intermediate ones. Video delay is a fact.
So you need to find a way to add the same delay to audio.

I don't know why it is not important when play file from disk and important when capture video and process frames on-the-fly.
Maybe answer is hidden here.

Audio track not processed by SVP. It is even not used in AVS script. Audio data is bypassing from ffdShow and Avisynth. So you need separate solution for audio.

This is something I have also battled with in my setup. Its difficult to get a svp threads + ffdshow buffer value that doesn't cause video desync. In my opinion because SVP is adding frames and can cause these delays, it should have the potential to correct any delay it adds by changing the frames timestamps. Or do you think its ffdshow / avisynth itself thats at fault?

Things like LAV filters will sync audio to video no matter how much the video gets delayed, once the video timestamps are correct coming out of svp / avisynth / ffdshow, but in some situations as described above, the video timestamps must be getting set very very badly wrong? Is this correct Chainik, MAG79? Is it possible to see on your end of things whether situations like desribed above produces timestamps that are forward or backward from where they should be?

No I don't think audio has to be touch, just if video gets delayed, the video frames timestamps should be updated to indicate that. The splitter like LAV splitter should be able to handle any delay in audio / video and still sync them together, once the timestamps are correct.

I assume in this case the video timestamps are being broken somewhere.

hmm, what ties it all together at the end into the final output, I guess its that components job to take audio / video and sync them together. Is it saving their timestamps in the captured output? If so again it will be up to the player to understand those and it will keep them in sync if the timestamps are right.

Thats my understanding anyways. I might try to see can I find an application that outputs timestamps of frames without avisynth / svp, and with, and see what happens to them, ie are they all offset by a certain amount depending on the buffering / threads.

Thats what I assume is happening. I'm hoping someone like Chainik or Mag79 might have tools to be able to see if this is happening or if SVP is in control of the timestamps. (I assume it is if its creating new frames).