Updated 12/14/22: Adding directions for v 1.03 and GUI tutorials.

In my eternal quest for interesting software to improve microphone quality of life I have uncovered an interesting gem named RNNoise, with a VST named Real-time Noise Suppression.

As the name indicates, it does roughly the same thing as programs like RTX Voice or Krisp, except it's free, works on any system (Including Mac and Linux), and so far appears to have few downsides on system performance.

I did a quick recording comparing no suppression vs. Real-time Noise Suppression vs. RTX Voice for both white noise and keyboard noise. The results are quite surprising. Give it a listen:

While obviously your voice gets butchered when a speaker is blasting white noise directly into the mic, at a still noisy 50% volume it's doing pretty well. I believe any relatively standard background noise will be repressed with minimal impact on audio quality.

Pros & Cons

	RTX Voice	RNNoise
Ease of Setup	Easy	Moderate - Hard
OS Compatibility	Windows Only	Win / Mac / Linux
Hardware Compatibility	Specific Nvidia Cards	Everything?!
Noise Reduction Quality	Excellent	Good
Bugs & Issues	Some	None

 

In short, I think RTX Voice performs slightly better on the quality front, and it is significantly easier to install, but it is less compatible and the bugs I found in my latest review hold it back from being suggested even if you have the hardware and OS to support it.

So how can you get this VST to work with your system? It's as easy as heading over to Github to download it and follow the directions below the download based on your OS.

https://github.com/werman/noise-suppression-for-voice

Then follow these directions

Windows:

Download EQ APO here: https://sourceforge.net/projects/equalizerapo/

Unzip, and start the installer for Equalizer APO. You should be greeted with the "Configurator" setup screen, which looks like this:

Select the "Capture Devices" tab and check the box for the microphone(s) you want to use. You do not need to select any Playback Devices.

 

Reboot. This step is mandatory, so don't skip it you lazy bums!

 

Open "Configuration Editor" from EQ APO. If you don't remember where you put EQ APO, you can just type this into the search bar and it should come up. It will look something like this:

Remove all the tabs that exist by pressing the red minus (-) sign.

Press the green plus (+) icon and add the following:

• Control -> Device
• Plugins -> VST Plugin
• Basic Filters -> Preamp

Set up each as follows:

Device- Select the mic you want to use. You can and should also select the device from the top drop down menu as seen below:

 

VST- Click the folder button as seen below and navigate to where you unzipped  RNNoise. Select the folder VST and the file "rnnoise_stereo.dll". When complete it should look similar to this:

 

Preamp- You do not need to adjust this, but you can make your mic louder or quieter by adjusting this number. This modifier will let you make your mic louder than "100%" on Windows or fine tune your volume to a much greater degree than any Windows setting. Great if your mic is too quiet (or too loud).

That's all you need to do! You can test to see if the noise reduction is working by recording yourself with background noise and pressing the white power button on the VST Plugin. You should hear a significant change when it is off vs. on.

EQ APO should automatically start when you reboot from now on and automatically reduce your background noise. If you notice it stops working, odds are it has forgotten the device (possibly because you unplugged it at some point), just re-open "Configuration Editor" and check the selected device to make sure it still has the correct mic selected.

 

New UI & Settings- In the most recent update they added several new features and UI elements, which we will cover below. You do not need to adjust these values, but in case you want to fine tune your noise reduction here's what they do:

To access the UI menu press the "Open Panel" button located in EQ APO's VST Plugin area, as seen here:

This should bring up a panel with 3 options:

VAD (Voice Activity Detection) Threshold- When you are not talking RNNoise acts as a complete noise gate, essentially muting your mic. This number adjusts how sensitive it is to deciding if you're talking or not. The higher the number, the more aggressive it is, so increase this number if it's picking up noises when you are not talking. Even at 95% it may pick up very loud noises, just fyi. It's science, not magic, so there's some limitations here!

VAD Grace Period- How long after the last voice detection the output won't be silenced. In short, if it is cutting off the final letters of your words, increase this grace period. I suggest small increments of 1-2 units as you test, try to keep this number as low as possible for the best results.

Retroactive VAD Grace Period- This is the same as the above but in reverse, if you are having the first sounds of your voice cut out, add values here until the problem resolves BUT keep in mind this feature adds latency to your voice. E.g. if you set this to 100ms, your voice will be transmitted 100ms after you talk.

 

Linux:

Follow our guide: https://antlionaudio.com/blogs/news/real-time-microphone-noise-reduction-linux-guide

Mac:

Follow the directions here: https://github.com/werman/noise-suppression-for-voice

 

If you run into this EQ APO error:

"The library is not readable by the audio service. Change the file permissions or copy the file to the VSTPlugins directory."

If you get this error don't panic, all is not lost. To fix this simply open up the folder and find the the rnnoise_stereo.dll. Right click on it and go to properties, then to the Security tab and click Edit.

In the next window click Add... and in the box type "LOCAL SERVICE" (No quotes) and give it read & execute access.

Then, just because it always makes me feel better to do so, reboot your machine.

As always you can get help with this process and any technical problem you may be having with your ModMic on our Discord: https://discord.gg/AntlionAudio