OBS Mic Test: Check Your Microphone Audio Before You Go Live

Nothing ruins a stream faster than a muted or clipping microphone, and this online OBS mic test catches both before you go live. It reads your mic in the browser with no download and shows the exact dBFS levels OBS Studio will receive, so a one-minute OBS audio test here tells you whether to raise gain, add filters or fix a setting before your next recording.

How to Run an OBS Mic Test in Your Browser

The tool uses the browser's Web Audio API, so your voice is analysed locally and never uploaded. Nothing is stored once you close the tab. It works the same on Windows, macOS and Linux, and with any device the operating system already recognises. Use the tv and soundbar test tool whenever you want a fast, repeatable way to confirm this.

1. Click Start and allow microphone permission when prompted.
2. Pick the same device you use in OBS from the input list.
3. Talk at your normal streaming volume for about 20 seconds while watching the live waveform and meters.
4. Note the average speech level, the loudest peaks and the level of the room when you are silent.
5. Read the frequency response view to see whether low rumble or harsh highs dominate your voice.

Because the OBS mic test tool reads the raw input before any OBS filter touches it, the numbers it gives you are your starting point for every setting below.

What the Microphone Tester Reads From Your Input Level

Every reading is in dBFS (decibels relative to full scale), where 0 dBFS is the digital ceiling and everything else is negative. The tester reports average speech, peak level, the noise floor of your room, any clipping, the device sample rate (usually 48 kHz) and a rough latency figure. Your input level is the number that drives every later decision, so get it right before touching filters.

Tuning Audio Filters With the Microphone Tester

Right-click your mic in the mixer and choose Filters. Order matters, because each filter processes what the previous one hands it. A clean filter chain for voice:

1. High pass filter set just under your voice's lowest note removes desk rumble and air-conditioning hum.
2. Noise suppression: RNNoise handles fans, hiss and keyboard noise more naturally; Speex uses less CPU on older machines.
3. Noise gate closes the mic between sentences.
4. EQ (equaliser): a small presence lift in the upper mids adds clarity.
5. Compressor evens out loud and quiet words so your voice stays above game audio.
6. Gain filter last, as a final trim.

Rerun the test after each change. If the room level falls well under -60 dB once suppression is on, the background noise is handled; if your voice starts sounding watery, lower the suppression strength. Hardware noise cancellation on a wireless kit can do the same job, so avoid stacking two aggressive stages of these audio filters.

Setting Noise Gate Thresholds From Room Noise

Set the noise gate's close threshold about 10 dB above your measured room level and the open threshold about 6 dB above that. In the example, with the room at -50.6 dB after gain, that is roughly -40 dB to close and -34 dB to open. Keep attack short (a few ms), hold around 150-250 ms so words don't get chopped, and a release near 100-150 ms. If your voice cuts in and out, the close threshold is too high; lower it by 3 dB at a time.

Checking a Speedrun Commentary Mic With a Test Recording

The decision here is simple: can one gain setting cover both calm commentary and the shout when a split lands, or does the chain need compression? The answer, worked out below, is that gain alone fails, and a 3:1 compression stage plus 3 dB of make-up gain fixes it.

A dynamic mic runs into an interface with the preamp at about two-thirds. I run the microphone test on this page twice: once talking through a route at normal volume, once shouting the way I do on a personal best. The readings come back as:

• Normal speech: -26.8 dBFS average, with louder words reaching -15.9 dB.
• The shout: -4.7 dB.

With only the calm pass, the gain window looks easy: speech needs at least +8.8 dB to reach -18, and normal peaks allow up to +9.9 dB before -6. So I try +9 dB. Speech lands at -17.8 and normal peaks at -6.9, but the shout goes to +4.3 dB, over the 0 dBFS ceiling, and that clips. The shout's own limit is -1.3 dB of gain, so no positive gain satisfies both readings. That conflict is invisible if you only test your normal voice.

So I keep +9 dB and add compression with its threshold at -20 dB and a 3:1 ratio, then 3 dB of make-up. By hand, anything above -20 keeps a third of its overshoot: the shout becomes -20 + 24.3 / 3 = -11.9, plus 3 = -8.9 dB.

I then make a short test recording with the shout in it and watch the OBS meter: the shout stays in the yellow. One caveat: these figures use a static curve. Real compression needs a few milliseconds to react, so the first instant of a shout can still overshoot. I'd add a limiter at -6 dB as a safety net rather than trust the arithmetic alone.

The input that matters most here is not the speech average but the gap between calm talk and the loudest moment, 22.1 dB in this case. Once that gap passes about 12 dB (the distance from -18 to -6), no gain setting alone can work.

The Pre-Stream Test: Run a Test Stream on Twitch or YouTube

The last pre-stream test checks the real output. On Twitch, add ?bandwidthtest=true to your stream key so the broadcast goes nowhere public; on YouTube, set the stream to Private or Unlisted. Start the test stream, then listen on a phone before you go live publicly, not on the PC running OBS, to avoid feedback and confused routing. If the voice is thin or quiet there but fine locally, check that the audio bitrate is at least 160 Kbps under Settings > Output.

Choosing an OBS Microphone for Streaming

Any of these works in OBS Studio, but your mic type changes how hard the filters have to work:

• USB mic (usually a condenser): plug and play, very sensitive, picks up the room.
• XLR dynamic mic through an audio interface: rejects room noise but needs plenty of gain; condensers on XLR need 48V phantom power.
• Headset mic: convenient, but use stronger RNNoise suppression and a tighter gate.
• Lavalier or wireless microphone: check the receiver shows up as an input in the OS and that the transmitter isn't muted before blaming OBS.

Mic Placement and Room Acoustics

Keep the mic about a hand's width from your mouth, slightly off-axis, and use a pop filter to stop plosive bursts. Soft furnishings cut the reverb that no filter fully removes. After moving the mic or softening the room, rerun the test: the silent-room reading should drop, and P and B sounds should stop pushing peaks toward -6 dB, which leaves your OBS gate and gain settings less work to do.