Question / Help Audio Filter Strategy For Large Number of Audio Sources

[jkao]

I'm running a stream while I am taking about 23 RTMP streams as input (from DJs streaming to me also using OBS via Media Sources) and I'm mixing them to a single output stream.

To ensure consistent audio, I have compressors and limiters on the input stream. The challenge I'm having is that as I adjust settings, I need to constantly be deleting filters and then doing "Copy Filters" from the source I modified filters on to the other 22 streams. I also can't use more advanced VST plugin as filters because I would have to instantiate them 23 times, configure them 23 times, and potentially run out of memory and CPU by having so many VSTs instantiated.

In a DAW, I would usually do this use case by routing the 23 source channels to a single output channel and applying filters just on the output channel, but the OBS audio mixer seems to only let me access a filter chain for input sources, and not a master output "source".

Does anyone have recommendation for strategies to manage the filter setup and adjustment of a large number of audio sources like this?

 

[Dapperstache]

Route them to RTX Voice and use that as your source?

 

[jkao]

Hmm.. when you say "route them", are you saying that there's a way to do source-to-source audio routing in OBS? Or suggesting doing something like turning on all those sources to Monitor Only output and setting the monitor device to a virtual loopback device, like Virtual Audio Cable or Voicemeter Banana, that will then show up as a Mic?

If the latter, I guess then you sacrifice local monitoring, since the "Listen to this Device" setting in the control will be prior to the filter in OBS and there's only one audio Monitor output?

That's an interesting idea, some trade-offs in monitoring, but plausible to set up.

 

[Shinotama]

Hmm.. when you say "route them", are you saying that there's a way to do source-to-source audio routing in OBS? Or suggesting doing something like turning on all those sources to Monitor Only output and setting the monitor device to a virtual loopback device, like Virtual Audio Cable or Voicemeter Banana, that will then show up as a Mic?

If the latter, I guess then you sacrifice local monitoring, since the "Listen to this Device" setting in the control will be prior to the filter in OBS and there's only one audio Monitor output?

That's an interesting idea, some trade-offs in monitoring, but plausible to set up.

Well I have all audio in-coming through Voice Meeter Potato, monitored there, then output via the VAIO options on there so I can split all my audio down into channels for OBS.

So I have RTX Audio using my Yeti for noise removal, outputting to VAIO AUX (B2), All Music I.e. Browser Audio for Music / Videos and Nightbot Via the Desktop app outputs to VAIO3 (B3), and Comms Such as Discord or Facebook Messenger Calls etc through to VAIO (B1) so that I can lower the audio dependant of what I require. In addition I also have the Desktop audio on it's own "Main Audio" channel and because I'm tired of Nightbot being crazy loud in my ear but not very loud on stream I've set it to playout to a Cable-D so I can self-monitor out to my headset / speakers.

 

[jkao]

Yeah, if I were using a bank of microphones, I could route that way. Unfortunately, since my 23 inputs are all live RTMP OBS Media Sources, they aren't individually available as devices outside of OBS. The only way to get the audio out of OBS that I can think of is to utilize the Monitor Output to route the combined audio out and then route it back in again, as a Mic/Aux source (but unmonitorable since I've already used up the Monitor Output).

I guess in theory, if I wanted to do all the filtering using VSTs, and not use any OBS audio filters, I could route audio out of OBS using Monitor Output, then use Voicemeter to route that into a DAW. Apply VSTs in the DAW, then route that back out through Voicemeter into both a virtual output for OBS to use as a Mic/Aux input, and to an audio device for monitoring. Then do all adjustments in the DAW and leave OBS with no filters and no gain adjustment that would make the monitoring at the Voicemeter level inaccurate.