dynamic audio normalizer filter?

[hedgehog90]

I am preparing for a gaming stream whereby game audio and voices (via Discord) are separate input sources.

In my testing, I've noticed some people have quieter mics than others, and if they're not talking over one another I want the people with quieter mics to be boosted.
In FFMPEG there is a helpful filter (dynaudnorm) and it achieves the desired effect perfectly.
But in OBS I'm struggling to get anything sounding half as good with the basic filters.
I tried a VST 2 filter called DynamicAudioNormalizer. It works but for some reason I get a 10-15 second delay on the audio when I have it applied and I can't get the VST interface to load, so I deemed it incompatible with OBS.
The best I can manage with basic filters is to use a compressor with an arbitrary amount of output gain, followed by a limiter:

(Ignore the 'invisible' filters)




...But this makes the loud parts of the Discord audio (people with good mic levels) sound boomy as a result.

I also have it setup so the gameplay audio ducks down low when people are speaking (using a Compressor with Ducking Source pointing to Discord), but this doesn't solve the problem regarding a group of mics with varying levels as 1 input (aka Discord).

Any ideas?

 

[Deleted member 121471]

I'm no audio expert but I can share what worked for me.

15:1 ratio and -10dB threshold on compressor while setting the limiter to only -8dB are pretty odd values, even without knowing any information about the input sources. You're making the loudest noises quieter but 15:1 is incredibly aggressive.

You could keep trying with the stock compressor, though I wouldn't use anything higher than 4:1, instead testing different thresholds, starting from -32dB and testing from there until it sounds more balanced.

Or you could be lazy and try Loudmax plugin. It's not perfect but I can't think of anything easier to solve your problem without introducing too many additional issues.

 

[hedgehog90]

No matter what settings I try on the compressor it just doesn't do what I need. Quiet voices remain quiet relative to louder voices, no matter what I set the ratio or threshold to, and the overall effect generally was very subtle. The settings in the screenshot are admittedly arbitrary except for the output gain.
However, I've just tried the Loudmax VST plugin you recommended and it does a much better job than my previous gain/limiter chain. I've set the threshold low so a quiet mic sounds no different to a loud mic on the same device, exactly what I was looking for! I've also capped the output to -10dB so it never goes into the red.

Before & After Loudmax:

Cheers!

 

[Game Maker George]

is er ook een VST plug-in voor OBS dat je het zelfde kan als audacity. Dus met OBS kun je het zelfde als audacity.


ik zoek namelijk

Een Normaliseer voor mijn microfoon. Die de audio even hard maakt

En een goede noice reduction. VST Plug-in. Die de ruis weghaalt.

Want ik wil ook goed geluid hebben tijdens het live streamen.







is there also a VST plug-in for OBS that can do the same as audacity. So with OBS you can do the same as audacity.


because I am looking

A Normalize for my microphone. Which makes the audio equally loud

And good noise reduction. VST Plugin. Which removes the noise.

Because I also want good sound during live streaming.

 

[AaronD]

is er ook een VST plug-in voor OBS dat je het zelfde kan als audacity. Dus met OBS kun je het zelfde als audacity.


ik zoek namelijk

Een Normaliseer voor mijn microfoon. Die de audio even hard maakt

En een goede noice reduction. VST Plug-in. Die de ruis weghaalt.

Want ik wil ook goed geluid hebben tijdens het live streamen.







is there also a VST plug-in for OBS that can do the same as audacity. So with OBS you can do the same as audacity.


because I am looking

A Normalize for my microphone. Which makes the audio equally loud

And good noise reduction. VST Plugin. Which removes the noise.

Because I also want good sound during live streaming.

Normalizing only works for a recording. It looks at the entire file, finds the largest peak, calculates a single gain to put that peak at the level that you specify, and then applies that single gain to the entire file. You can't do that live, because you don't know what's coming.

For live work, the best you can do is a compressor. OBS already has a filter for that, no need for VST. The posts above must not know how to use one because it does work. Lots of tutorials on YouTube; look them up, without being specific to OBS. Understand the general use, and then translate that back to OBS.

For noise reduction, OBS already has that too. Again, no need for VST.
However! Make the room as quiet as possible before you add the noise suppressor! It's not magic. Turn off the fridge, HVAC and other fans, phone ringer, etc. The less you make it do, the better the result will be.