Question / Help Separating multiple NDI user audio?

[plzhalp]

Not sure if I'm overcomplicating this: I'm patching in two co-workers via Skype video conference (they're on separate machines, separate locations). However their audio apparently goes thru one NDI Source audio channel on OBS and echoes terribly.

Is my best bet to bone up on Virtual Audio Cable (I'm a converted Mac/Loopback guy) and figure out how to program Skype properly so that I have them on separate channels in OBS? Is that possible?

I actually just experimented and enabled "Desktop Audio" which puts both of them on one audio channel in OBS and neither echoes, so that's encouraging; I meanwhile am on my own separate "Audio Input Device" channel. My concern is some lip sync issues and not being able to adjust their volume individually but sometimes close enough is fine by me if VAC is a complex thing.

Opinions and such are appreciated.

 

[Narcogen]

As far as I understand in Skype this is not possible. This is a flaw in their implementation of NDI that they are not providing separate isolated audio/video sources.

 

[plzhalp]

As far as I understand in Skype this is not possible. This is a flaw in their implementation of NDI that they are not providing separate isolated audio/video sources.

Well good to know it's not operator error, ha. Thanks!

Just got done with a stream on YouTube, unfortunately the lip sync was pretty bad; audio is several frames ahead of the lips on the Skype guys. Is there any tweak I can do on OBS to help compensate -- changing the sample rate in the Audio tab or the audio bitrate on the Output tab? YouTube only has the ultra-low, low and normal latency settings (stream was performed on 'low', think normal bogged my bandwidth and crashed the stream).

 

[Narcogen]

If the audio is ahead you can add delay in Edit > Advanced Audio Properties. This is per-device, though.

 

[plzhalp]

Excellent. I'll try that out. Thanks for the help!

 

[plzhalp]

If the audio is ahead you can add delay in Edit > Advanced Audio Properties. This is per-device, though.

I added a 54 ms delay on the 'Desktop Audio' but didn't notice much of a difference on the lip sync -- should I have put the delay on the NDI sources instead?

 

[Narcogen]

NDI sources are audio and video. If you're trying to affect lip sync, yes, you should be adding the audio delay to the sources that contain the Skype video-- the NDI sources.

Desktop audio is sound from other applications.

 

[plzhalp]

NDI sources are audio and video. If you're trying to affect lip sync, yes, you should be adding the audio delay to the sources that contain the Skype video-- the NDI sources.

Desktop audio is sound from other applications.

Got it. Thanks, again!

 

[plzhalp]

Hey @Narcogen this worked out well in one aspect -- audio was synched up perfectly for the Skype guest via NDI. However, despite waiting 1-2 seconds after he was done talking to ask another question, it sounded like I was interrupting him every time. Do I need to put some sort of delay on myself as well so I'm not stepping over him?

 

[Narcogen]

If you're grabbing your own audio locally, then yes, OBS will be getting your audio before your participant does, so adding some delay might help this effect.

 

[plzhalp]

If you're grabbing your own audio locally, then yes, OBS will be getting your audio before your participant does, so adding some delay might help this effect.

I do grab my audio locally. Will this then create a lip sync issue with my video and audio? Are there any other options, hardware or software, to compensate or should I just use 'Desktop Audio' for everyone on the stream and adjust the delay on the video sources? Also maybe just easiest to just tell myself to wait 2-3 seconds before speaking once he's done?

 

[Narcogen]

Any remote participant is going to be delayed by some amount.

If you're got a local cam and a local mic then you can't delay your audio to better mesh with the audio of the remote participants unless you also delay your camera video, but you can do both.

 

[plzhalp]

Any remote participant is going to be delayed by some amount.

If you're got a local cam and a local mic then you can't delay your audio to better mesh with the audio of the remote participants unless you also delay your camera video, but you can do both.

Things have been working smoothly when I patch in one guest via NDI -- no sync issues, no inadvertent interrupting. However when I bring two guests in, the lip sync issue rears its head again. Strange thing is, I'll press 'start recording' for a mic/sync check beforehand and everyone is good but once I hit 'start streaming' one of my guests will become out of sync quite quickly...and sometimes it'll go away and then the other guest starts going out of sync.

This is my set up: I'm on the Logitech cam (video only) and my audio is the "Mic/Aux", one of my guests will under "Corey", the other as "NDI Source". Any theories?

 

[Narcogen]

Are your encoder settings for recording and streaming the same-- are you recording with the stream encoder settings?

If not, that sounds to me like your streaming settings are creating greater load than your recording settings (which would not be unusual as streaming is usually more bitrate restrained, and therefore requires more effort to reach reasonable quality levels).