Video and audio synchronization problem

[straannick]

Hello! I need to sync audio and video for an OBS broadcast at 2560x1440, 30 fps with the following configuration:

• Audio: FocusRite 2i2, 3rd

Video, 4K webcams:

• LiteTrack in 3840x2160 mode
• Logitech Brio in 3840x2160 mode
• AverMedia PW513 in 3264x1836 mode. It doesn't have a 2560x1440 mode, and it has the highest signal lag, so the other two use 3840x2160.

I conduct test broadcasts, then download them and load video into a video editor, where I test the audio and video sync.

Through trial and error, I finally achieved the following result:

LiteTrack is approximately 8.5 frames ahead of the audio, Brio is 8.5 frames ahead of the audio, and AverMedia is 8.5 frames ahead of the audio. The latencies in Advanced Audio Properties are as follows:
LiteTrack -924 ms, Brio - 1188 ms, AverMedia - 0 ms, FocusRite - 693 ms

Excellent! So, all that's left is to reduce FocusRite's latency by 8-9 frames, or (1000/30)*8.5 ms.
I create a new broadcast with latency values of:
LiteTrack -924 ms, Brio - 1188 ms, AverMedia - 0 ms, FocusRite - 420 ms.

And I get the following unexpected result:
LiteTrack is lagging by about 8 frames, Brio is lagging by 3.5 frames, and AverMedia is ahead of the audio by 1.5 frames. How can this be?!

So I revert to the previous latency values, where all cameras were ahead of the audio by the same amount – 8.5 frames – expecting a similar result. However, in the new broadcast, I'm getting the following results:
LiteTrack is lagging behind by about 3.5 frames, Brio is ahead by 4 frames, and AverMedia is ahead by 2.5 frames—that is,
completely different from what I saw previously with similar delays.

Can anyone explain why this is happening and how to achieve audio and video sync?

Thanks!

 

[straannick]

Addendum. After another experiment, I achieved synchronization FocusRite audio with Aver Media and Brio at certain latency values. I then decided to achieve overall synchronization by adjusting only the Lite Track latency. Take a look at the results I obtained. I've compiled them in a table for clarity. To date, I've conducted 36 test broadcasts and have not achieved any positive results! So far, the obvious conclusion is that latency adjustment in OBS is flawed.
If I'm wrong, please explain - why!

 

[straannick]

A new series of experiments.

The previous experiments began with latency values obtained by synchronizing (quite successfully in previous versions of OBS) the same webcams with a MixPre 6 II audio source. Now I decided to start from scratch, setting all latency values to 0. Initially, I ran two parallel broadcasts using the SoraYuki plugin, but starting with experiment #7, I switched to only one broadcast. I also tried not to change multiple latency values simultaneously.

The table presents the results, which allow us to draw several conclusions:

• Latency values for different audio/video sources affect the behavior of other sources;

• The relationship between latency values settings and the actual behavior of signal sources is not linear (not proportional to the number of frames), so there is no point in taking into account the actual frame duration (1000/30 ms);

• Different broadcasts with the same latency settings will not have the same actual signal latency (synchronization is unstable);

• Absurd results are possible. For example, increasing the audio source delay (FocusRite) may increase, rather than decrease, the actual video delay.

• The number of parallel broadcasts using the SoraYuki plugin significantly affects the behavior of signal sources.

I can provide links for all broadcasts in the table.

The overall conclusion is that this very useful feature of signal synchronization is practically useless in the current version of OBS!

I would like technical support to comment on the results and conclusions obtained and explain how to synchronize video and audio from different signal sources and ensure stable synchronization from broadcast to broadcast!