Optimal Compressor Settings for Stream Mic: Best Settings and Config (2026)

If your stream mic sounds either too quiet when you whisper or painfully harsh when you laugh and shout, you need compression before you need a more expensive microphone. A compressor narrows your dynamic range so your quietest intelligible speech stays above your background noise and your loudest peaks stay below clipping and harsh limiting. For streaming, the priority is not loudness for its own sake, it is consistency and intelligibility with minimal pumping or breathing artifacts. That means relatively moderate ratios, a threshold set just where your normal voice sits, timing constants that catch plosives without crushing transients, and makeup gain applied only after the other controls are dialed in. The settings below work in OBS Studio, Streamlabs Desktop, NVIDIA Broadcast, VoiceMeeter, and most VST hosts, and reflect the current best practices for voice in 2026.

Quick Recommended Settings

Setting-by-Setting

Threshold: -18 dB to -14 dB

Set this so the compressor rests at 0 dB of gain reduction when you are silent and applies 3 to 6 dB of reduction when you speak at your normal streaming volume. In OBS Studio right-click your mic in Audio Mixer > Filters > + under Audio Filters > Compressor > move the Threshold slider while watching the gain reduction meter. In Reaper or any VST host using ReaComp, the threshold is the horizontal slider labeled Threshold. If you set it too high at -10 dB, only shouts are compressed and the level will still wander. If you set it too low at -30 dB, even typing and breathing will pump the gain. The trade-off is sensitivity versus transparency: lower catches more variation but risks noise pumping.

Ratio: 3:1 to 4:1

This controls how much the signal is reduced once it passes the threshold. At 3:1, for every 3 dB over threshold, only 1 dB passes through. This range is the sweet spot for live voice because it evens out 6 to 9 dB of performance dynamics without flattening emotion. In OBS the Ratio slider is directly below Threshold. In NVIDIA Broadcast > Microphone > Voice Filter > Compressor > Ratio sets the same value. Pushing to 6:1 or higher gives a broadcast-crushed feel that works for loud esports comms but quickly sounds unnatural on a solo stream. Staying below 2:1 often is not enough to keep whispers audible and laughs controlled in the same sentence.

Attack: 10 ms to 20 ms

Attack is how quickly compression engages after crossing threshold. For a stream mic you want plosives like P and B and sudden laughs caught fast, but not so fast that you shave off the initial transient of every word. At 10 ms to 20 ms you preserve clarity on words like kick and take while still taming spikes. In OBS the Attack is labeled in milliseconds. In VoiceMeeter Banana > Compressor knob > right-click for Expert Mode exposes Attack in ms. Use 5 ms or less only if your mic is very peaky up close. Use 30 ms or more if you want more punch for an ASMR or narration style voice where transients are part of the character. Faster attack equals more control but less natural attack.

Release: 80 ms to 140 ms

Release sets how fast gain returns to normal after level drops below threshold. Too fast at 20 to 40 ms creates audible pumping as the compressor releases between syllables. Too slow at 300 ms or more holds gain down and makes the start of the next phrase dull. Set release so gain reduction returns to zero in the natural pause between sentences. In OBS move Release while speaking the test phrase testing one two three and watch the reduction meter settle. In FabFilter Pro-C 2 or TDR Kotelnikov, choose the dedicated Release dial. The trade-off is smoothness versus responsiveness: longer release is smoother but can suppress quiet word beginnings.

Knee: 6 dB Soft Knee

Knee determines how gradually compression begins around the threshold. A hard knee at 0 dB applies the full ratio immediately, which can sound abrupt on voice. A soft knee of 6 dB introduces compression gradually over a 6 dB window, so a speaker whose level wanders a few dB does not trigger obvious on-off compression. In ReaComp select Soft Knee and set Knee size to 6 dB. In OBS Studio’s stock compressor the knee is fixed to soft, so no action is needed beyond understanding its behavior. If your compressor offers a knee control and you are on a noisy keyboard, a harder knee can help keep compression from creeping in on light input, at the cost of more noticeable transitions.

Makeup Gain / Output Gain: +3 dB to +6 dB

Compression reduces peaks, so you add makeup gain to bring average loudness back up. Add it last. After you have 3 to 6 dB of gain reduction on normal speech, raise Output Gain until your loud peaks hit -6 dB to -3 dB on the OBS mixer and your quiet speech still sits above -24 dB. Key combo to monitor accurately: in OBS press Alt + drag mixer to expand meters, or open View > Docks > Stats to see peak values. Over-applying makeup will push self-noise and room hiss up, so if you need more than +8 dB, lower your threshold by 2 dB instead and re-check. The trade-off is loudness and presence versus noise floor.

Detector High-Pass Filter: 80 Hz to 120 Hz

This is applied to the sidechain, not the audible signal, so it does not thin your voice. It prevents low-frequency energy from dominating the detector. Turn on Sidechain HPF or Detector HP and set to 80 Hz for most boom mics, 120 Hz if you have desk rumble or a mic without a built-in low cut. Location: OBS Compressor > Sidechain/High-pass Filter checkbox then dial. In Equalizer APO + ReaPlugs, enable HP in ReaComp’s Detector input. Without it, a table bump or proximity thump can trigger 6 dB of reduction and duck your voice noticeably. The trade-off is immunity to rumble versus failing to catch low vocal energy if you set it too high above 200 Hz.

By Hardware Tier

Low Tier: Laptop Mic, Headset Mic, $20 to $50 USB

These mics have higher self-noise and pick up more room. Apply noise suppression before compression so the compressor does not ride on hiss. Path: OBS > Filters for your mic > add Noise Suppression with RNNoise > then add Noise Gate with Close Threshold -40 dB and Open Threshold -36 dB > then add Compressor. Use the gentler end of the range: Threshold -16 dB, Ratio 3:1, Attack 20 ms, Release 120 ms, Output Gain +3 dB. Keep makeup conservative to avoid raising hiss. File location for OBS profiles if you want to back up before changes: %appdata%\obs-studio\basic\profiles. Use Duplicate filter set by right-clicking Filters > Copy and Paste Filters to A/B quickly with key combo Ctrl + Z to undo a filter change.

Mid Tier: USB Condenser or Dynamic Like Solo, NT-USB, MV7 USB, Yeti – $80 to $220

This is where the quick recommended table works as written. Place mic 6 to 8 inches away, slightly off-axis, add a pop filter. Chain: Noise Gate at -38 dB close / -34 dB open > Compressor at Threshold -18 dB to -14 dB, Ratio 3:1 to 4:1, Attack 15 ms, Release 100 ms, Knee soft, HPF 100 Hz > Limiter at -2.5 dB ceiling as safety net. In OBS the Limiter is Filters > + > Limiter > Threshold -2.5 dB. In Streamlabs Desktop the path is Mixer gear icon > Filters > Compressor with identical names. This tier can handle +4 to +6 dB makeup cleanly and benefits most from a soft knee while keeping attack fast enough for game callouts.

High Tier: XLR Dynamic or Condenser via Interface or GoXLR, SM7B, SM58, RE20, LCT 440

With low self-noise and better preamps you can be more precise. Insert processing at the interface if it offers DSP to save CPU: for example, GoXLR App > Mic Setup > Compressor > set Threshold so gain reduction hovers 2 to 4 dB on normal speech. Otherwise use a VST chain in OBS with ReaComp or TDR Kotelnikov GE. Use Threshold -14 dB, Ratio 4:1, Attack 10 ms, Release 80 ms, Soft Knee 6 dB, Detector HPF 80 Hz, and 1.5 ms Lookahead if your plugin offers it. Follow with a Limiter at -1.5 dB. High-tier setups let you use slightly faster timing without pumping because the signal to noise ratio is better, which gives tighter control on loud laughs without muting breaths.

Common Mistakes

Cranking Ratio Instead of Fixing Threshold

What causes it: setting threshold too high so normal speech never triggers compression, then compensating with 8:1 or higher ratio to force leveling on peaks. How to check: speak at normal stream level and watch the gain reduction meter in OBS Filters > Compressor. If it shows 0 dB most of the time and then 10 dB on shouts, threshold is too high. What to do: lower Threshold by 4 dB increments to -18 dB, return Ratio to 3:1, then add +3 dB Output Gain. How to undo: in Filters click the eye icon to bypass Compressor or press Ctrl + Z to revert the last slider change and set Ratio back to 3:1.

Too Much Makeup Gain Raising Hiss and Keyboard Noise

What causes it: adding +10 dB or more makeup to sound loud after heavy compression. How to check: mute your game, go silent for 5 seconds with compressor enabled, and listen on headphones while watching OBS’s mic meter. If meter shows -50 dB to -40 dB when silent and noise is audible, makeup is too hot. What to do: lower Output Gain to +3 dB, raise Windows Sound > Input > Volume to 80 to 90, and use a Limiter at -2.5 dB instead of gain for loudness. Move Noise Suppression above Compressor in the filter chain. How to undo: drag Noise Suppression back below Compressor or reduce Output Gain by 3 dB and press Ctrl + Z if needed.

Fast Release Causing Pumping Between Words

What causes it: release at 20 to 40 ms that recovers before a word fully ends, creating audible breathing of volume. How to check: record a test with OBS > Controls > Start Recording, say phrases with pauses like hello … hello … hello, then playback and watch reduction meter bounce to 0 dB between syllables. What to do: increase Release to 100 ms to 140 ms and test again until meter returns smoothly during pauses between sentences rather than between syllables. How to undo: return Release to 100 ms default by resetting the slider or disabling the filter with the eye toggle.

Compressor Before Noise Gate

What causes it: placing Compressor above Noise Gate in OBS Filters list, so breathing and fan noise are amplified and then gated harshly. How to check: open OBS > Filters for mic source and look at order from top to bottom. Audio flows top to bottom. If Compressor is above Gate, order is wrong. What to do: click and drag Noise Gate and Noise Suppression above Compressor, so chain is Suppression > Gate > Compressor > Limiter. Test with keyboard typing while speaking. How to undo: drag filters back to previous order or right-click Filters > Remove on the moved filter and re-add it.

FAQ

Should I use the OBS stock compressor or a VST plugin?

Use the stock OBS Compressor if you want stability and tiny CPU load. It covers Threshold, Ratio, Attack, Release, and Output Gain with a fixed soft knee and optional sidechain HPF. Use a VST like ReaComp, TDR Kotelnikov, or FabFilter Pro-C 2 when you need features the stock lacks such as knee control, lookahead, RMS versus peak detection, dry/wet mix for parallel compression, or more accurate gain reduction metering. Install VST3 to C:\Program Files\Common Files\VST3 then OBS > Filters > + > VST 3.x Plugin > select the compressor. Keep one compressor active at a time to avoid double compression.

How do I know if my threshold is set correctly without guessing?

Open OBS mixer meter and the Compressor gain reduction meter together. Speak as you would on stream for 30 seconds including quiet asides and a loud laugh. Aim for 0 dB reduction in silence, 2 to 4 dB on normal sentences, and 6 to 8 dB on the loudest peaks. If you see consistent 7 dB or more on normal speech, raise threshold by 2 dB. If you see 0 to 1 dB on normal speech, lower threshold by 2 dB. Save a snapshot of your profile first via OBS > Profile > Duplicate for quick restore if you overshoot.

Do I still need a limiter after the compressor?

Yes, add a Limiter as the last filter after Compressor for safety. The compressor evens out performance dynamics, the limiter catches stray peaks and protects against digital clipping if you lean into the mic. Set OBS > Filters > + > Limiter > Threshold to -2.5 dB to -1.5 dB with Release 60 ms. This has almost no effect during normal speech but prevents red clipping during shouts or when a Discord sound blasts through. If you already use a hardware limiter in GoXLR or your interface, disable the software limiter to avoid stacking.

Why does my voice sound dull or squashed after compressing?

Dullness comes from too much gain reduction, too fast an attack, or excessive low-pass filtering elsewhere. First lower Ratio to 2.5:1 and raise Attack to 20 ms so consonant transients recover. Next raise Threshold by 2 to 3 dB so you reduce only peaks rather than the body of the voice. Finally check that you have not left an EQ low-pass at 8 kHz or enabled a warm voice preset that rolls off presence. To quickly compare, use OBS Filters eye icon to bypass the compressor and Limiter, record 15 seconds with Ctrl + R style workflow via Controls > Start Recording, then toggle and record again for direct A/B.