Which sound problem do you actually have?
Most poor stream audio comes down to one of five problems, and each has a different cure, so name yours before changing anything. Play back a recording of a past stream on headphones and match what you hear to a row below.
| What you hear | Likely cause | Where to fix it |
|---|---|---|
| Hiss, fan hum or air conditioning between sentences | Steady background noise reaching the mic | Mic placement first, then Noise Suppression |
| Crackle or harsh distortion on loud words | Clipping: the signal is louder than the gear can carry | Lower the gain on the device, then add a Limiter |
| Voice lost under music or video clips | Sources not balanced against each other | Mixer faders and a ducking compressor |
| Boomy, muddy voice | Mic very close (bass build-up) or a bare, hollow room | Move the mic back a little and soften the room |
| An echo of your own voice | Speakers playing back into the microphone | Headphones and the right monitoring choice |
Start with the microphone and the room
No filter can bring back a voice the microphone never captured cleanly, so the cheapest improvement is usually moving things around. Two words you will meet a lot: gain is how much the microphone's signal is amplified before anything else touches it, and clipping is the crunchy distortion you hear when a signal is too loud for the equipment carrying it.
- Use a directional microphone. Microphone maker Shure explains that directional mics pick up less room sound than omnidirectional ones, and that a cardioid mic is least sensitive at its rear. Aim that back side at the noisiest thing nearby: the computer, a window, the door.
- Get close, but not too close. Directional mics have the proximity effect, meaning the bass grows as you move nearer. If you sound boomy, back off a little and listen again.
- Soften the room. Bare walls and hard floors bounce your voice back into the mic. Curtains, a rug, a sofa or a full bookcase all help, and our home radio studio guide lists cheap ways to quieten a spare room.
- Wear headphones. If your speakers play the stream's sound, the microphone hears it a second time and viewers get an echo.
How loud should my microphone be?
Your speech should sit in the upper part of the yellow zone on the OBS meter, with only the loudest words touching the start of the red. The OBS Audio Mixer guide describes the red zone as close to clipping, puts game or video sound lower in the yellow, and keeps background music and alert sounds in the green.
- Find the gain control on the device itself: a knob on the microphone, the audio interface or the mixing desk, usually labelled Gain. The OBS guide says to always start at the device.
- Talk as loudly as you will at your liveliest moment (a laugh, a goal, the closing line of a sermon) and set the gain so that only those moments reach the red.
- Watch the small peak marker on the meter. It holds the loudest point of the last 20 seconds, so a marker sitting at the far right means something clipped.
- Only if the device has no gain left, add a Gain filter in OBS. Its guide treats it as a help for very quiet sources and says gain should generally be applied before the sound reaches OBS.
- Record a short session exactly as you would stream, then play the file back on headphones and on a phone speaker. If both sound clear and nothing crackles, your level is right.
Raising gain raises the room noise by the same amount. If you have to push it hard, move the microphone closer instead. And if your sound comes from a church or school sound desk, set that feed's level at the desk: a feed that arrives distorted stays distorted in OBS.
Add the four filters that do most of the work
OBS Studio has audio filters that run on each source separately. Open the menu next to your microphone in the Audio Mixer (or right-click the source), choose Filters, and add each one with the + button under Audio Filters. They run from top to bottom, so add them in this order.
- Noise Suppression removes steady background noise such as a computer fan. The default method, RNNoise, is higher quality but uses more CPU; Speex is adjustable with a Suppression Level slider (default -30 dB). The OBS guide warns that it will not rescue a loud room, and that strong settings can distort your voice too.
- Compressor turns down the loudest parts of your voice so soft and loud words end up closer together. OBS places it at or near the beginning of the chain. Its defaults are Ratio 10:1, Threshold -18 dB, Attack 6 ms and Release 60 ms; Output Gain brings the overall level back up afterwards.
- Expander lowers the level when you stop talking, so breaths and room noise fade between sentences. The OBS guide puts it after compression and before a limiter, calls a 4:1 ratio a nice balance, and recommends an attack of 5-10 ms and a release of 50-120 ms. Raise the Threshold from its default of -40 dB until the noise goes, and stop before the ends of your words get cut off.
- Limiter is a ceiling no sound can cross, which saves you from a sudden shout or cheer. Its default Threshold is -6 dB, and OBS says it should be the last filter in your chain.
Prefer a plain gate? The Noise Gate filter opens and shuts at two levels: put Close Threshold above the noise and Open Threshold a little below your voice (defaults -32 dB and -26 dB). A gate silences the sound completely, while an expander at a moderate ratio only turns it down, which tends to sound less abrupt.
Keep music underneath your voice
Background music belongs in the green part of the meter and should dip whenever you speak, and OBS can do that dipping for you. It is called ducking: a compressor on the music source listens to your microphone and lowers the music while you talk.
- Open the filters of the music or desktop audio source (not the microphone) and add a Compressor.
- Set Sidechain/Ducking Source to your microphone.
- Begin with the values the OBS guide suggests: Ratio 32:1, Threshold -36 dB, Attack 100 ms, Release 600 ms, Output Gain 0 dB.
- Talk over a song. The music should drop while you speak and rise again shortly after you stop.
- Move Threshold to make the dip deeper or gentler; Attack and Release set how quickly it falls and recovers.
When the music is the programme itself, such as a choir, a school concert or a band, leave ducking off for that source and balance it with the faders instead.
Hear what OBS hears without an echo
Use audio monitoring, and always through headphones. In Edit > Advanced Audio Properties, each source has an Audio Monitoring choice: Monitor Off, Monitor Only (mute output) or Monitoring Enabled (called Monitor and Output before OBS 33). Choose Monitoring Enabled on your microphone while you set up, then go back to Monitor Off if hearing yourself distracts you during the show. The same window lets you type a volume percentage well above 100% when a fader alone is not enough.
Test it on a real channel before the big day
A recording tells you how you sound on your own computer; a test broadcast tells you how you sound on a viewer's phone. Our live video channels take RTMP from OBS Studio, vMix, Wirecast, Larix or a hardware encoder and play over HTTPS/HLS in any browser. The Free plan is made for this kind of trial: one channel for up to 10 viewers at a time, an embeddable player without ads and help by email. Ask a friend to open the link on a phone at half volume and tell you whether every word comes through over the music.
If the sound stutters or cuts out rather than sounding poor, look at the connection and the encoder before the filters: upload speed for live streaming and OBS dropped frames walk through both.
Running an audio-only show? A radio station may fit better: the Pro radio plan adds a studio encoder in the browser, up to 256 kbps with sound processing.