How to clean up your microphone with noise removal, EQ and a compressor

The order of the effects

Every input strip has the same chain, in this order: noise removal, gate, EQ, compressor, then the fader and the output limit. Each effect can be switched off on its own, and the DSP view shows the chain above its cards so you can see what is on.

• Microphone
• Noise removalsteady noise
• Gatesilence between words
• EQtone
• Compressorlevel
• Fader + limit

Bring in your microphone

In the Mixer, choose Add input. Under Hardware inputs pick your microphone. If it has several channels, pick one: a single microphone on an interface is usually Mono · in 1.

Speak at your normal volume and watch the meter. Use the fader so loud words reach the upper part of the meter without touching red.

Open the effects

Every strip shows small chips for its effects: NR for noise removal, EQ, CMP for the compressor and GATE. A lit chip means the effect is on. Click a chip to open the DSP view for that strip.

Switch on noise removal

The noise removal card comes first. Turn on Enabled and talk. The Reduction meter shows how many decibels are being taken out, and the state reads Removing while it works and Quiet when there is nothing to remove.

Leave Strength at 100 percent to begin. If your voice sounds thin or watery, lower it until it sounds natural again.

Noise removal targets steady background noise such as fan hiss and hum. It adds about 20 ms of delay while it is on, and uses less than 1 percent of one CPU core per microphone channel. For quick sounds between your words, use the gate.

Set the gate

The gate turns the sound down when you are not speaking. Open the Gate card and speak while you watch its open and closed indicator.

1. Stay quiet and raise Threshold until the gate stays closed.
2. Speak softly, as you would at your quietest, and lower the threshold a little until the gate opens on every word.
3. Set Hold and Release so the ends of words are not cut off.
4. Use Range to decide how far the sound is turned down while the gate is closed. A little less than full often sounds more natural.

Shape the tone with EQ

Open the EQ view. Click Analyse my voice, read a few sentences aloud for about five seconds, and Soundweave suggests up to five bands and a high-pass filter. Use A/B to compare before and after, then apply. The analysis runs on your PC, and the recording is deleted when it finishes.

You can also start from a built-in preset, or drag bands on the graph. Band types are high-pass (12 and 24 dB per octave), low shelf, bell, high shelf, low-pass and notch. Save as keeps your own presets.

Even out the level with the compressor

The compressor makes loud words quieter so the whole sentence sits at a steadier level. Its card has Threshold, Ratio, Attack, Release, Knee and make-up gain, with a gain reduction meter.

1. Set Ratio between 2:1 and 4:1 for speech.
2. Lower Threshold until the gain reduction meter shows about 3 to 6 dB on your loudest words.
3. Raise make-up gain until the overall level matches how loud you were before.

Add a safety limit

The output limit is a ceiling the strip cannot go above, which protects whatever listens from a sudden shout. Click the LIM chip on the strip, switch the limit on, and set the ceiling a little below 0 dB, for example −3 dB.

Send the clean microphone to your apps

The effects are on the strip, so the apps need to listen to the strip’s output. Give them their own microphone:

1. On the Devices page, choose Add Cable and name it Mic.
2. In the mixer, choose Add bus, type B · virtual, name it Clean mic, and set its output to the Mic cable under Virtual outputs.
3. On your microphone strip, click the B button of that bus.
4. In Discord, OBS or your call app, choose Mic (Soundweave 1) as the microphone.

The OBS guide shows steps 1 and 2 with pictures.

Save it as a profile

A profile stores your strips, buses, routing, effects and hotkeys. Open Settings › Profiles and save, and you can come back to this setup in one click from the title bar.

Troubleshooting

My voice sounds thin or watery

Lower the noise removal Strength. If it is still thin, check that the EQ does not cut too much below 200 Hz.

The ends of words are cut off

Lower the gate Threshold, raise Hold or Release, or make the Range smaller.

The background noise pumps up after each word

Make the compressor Release longer and lower the make-up gain.

I hear myself a moment late

Noise removal adds about 20 ms while it is on. Use direct monitoring on your interface, if it has it, for what you hear in your own headphones.

The other app sounds quiet

Check the fader on the strip and the Clean mic bus, and the input volume of the app itself.

Questions

Keep going