How to Test Your Microphone: A Complete Checklist

Last updated 8 August 2026 — MicTest.live

"Testing your microphone" usually means checking that a bar moves. That confirms the device is alive and tells you nothing about how you will actually sound. This is the full checklist, in the order that finds problems fastest.

1. Does it produce any signal at all?

Open the microphone test, press start, allow access, and speak. If the meter moves, everything from the capsule to the browser is working, and you have eliminated hardware, cables, drivers and operating-system permissions in one step.

If it does not move, work through the device dropdown before assuming anything is broken. Computers routinely default to a disconnected headset or a monitor's built-in microphone.

2. Is the level right?

Speak at the volume you will actually use — people habitually test at a louder volume than they later speak at, which is why levels that looked fine turn out to be quiet in the real call.

You want peaks between −12 and −6 dB and an average near −20 dB. Below −40 dB average, listeners will strain. Touching 0 dB means clipping, which sounds harsh and cannot be repaired afterwards.

3. How loud is your silence?

This is the check almost nobody runs, and it predicts recording quality better than any other single number. Stop talking completely and watch the meter for ten seconds. That reading is your noise floor.

Noise floor	What it means
Below −55 dB	Excellent. Nothing to fix.
−55 to −45 dB	Good. Audible only in very quiet passages.
−45 to −35 dB	Noticeable in recordings. Worth improving.
Above −35 dB	Obvious to everyone. Fix the room before recording anything.

4. What do you actually sound like?

Record five seconds and listen with headphones. Not speakers — headphones, because they reveal detail that speakers in a room hide.

Listen specifically for four things. Hiss during the gaps, which means gain is fighting a weak signal. Echo, where your voice seems to arrive twice, meaning the room needs soft surfaces. Plosives, the thumps on P and B sounds, fixed by a pop filter or by speaking slightly across the microphone. Tone — whether you sound thin and distant, which means too far away, or boomy, which means too close.

5. Does it survive the processing?

Turn noise suppression and automatic gain control off using the toggles on the test page, and record the same five seconds again. Compare.

If the processed version is noticeably better, leave the processing on for calls. If the processed version sounds hollow, underwater, or cuts off the beginnings of words, turn it off and fix the underlying problem instead — the processing is doing damage to compensate for something you can address directly.

6. Does it work in the app you will actually use?

A microphone that works in a browser test can still fail in a specific application, because permission and device selection are handled separately in every app. Before anything important, open the actual application and run its own test: Zoom has Test Mic under Settings, Audio; Discord has "Let's Check" under Voice & Video; Teams has a test call.

Do this on the same machine, in the same room, on the same power state you will use for the real thing. Laptops behave differently on battery.

7. Latency, if you are monitoring yourself

If you will hear yourself in headphones while recording, check the delay. Anything above roughly 20 milliseconds is distracting; above 40 ms it becomes difficult to speak naturally, and for singing and instruments it is unusable.

USB microphones with a built-in headphone jack and audio interfaces both provide direct monitoring that bypasses the computer entirely, giving effectively zero latency. Bluetooth headsets are at the opposite extreme, typically 100–300 ms, and cannot be used for monitoring at all.

The pre-call checklist

Fifteen minutes before something that matters, run through this:

1. Plug in the laptop — battery mode changes both camera and audio behaviour.
2. Close every other app that could hold the microphone. Check the system tray.
3. Run the microphone test: level, noise floor, five-second recording.
4. Check the webcam in the light you will actually have.
5. Open the meeting app and run its own audio test.
6. Put headphones on. This alone prevents every echo problem.
7. Silence notifications, which are picked up by the microphone as well as heard by you.

Testing the noise floor properly

Everyone tests whether the microphone picks up their voice. Almost nobody tests what it picks up when they are silent, and that measurement predicts how you will sound far better than the first one.

Start the level meter, stop talking, and stay still for ten seconds. Whatever the meter settles at is your noise floor: the fan, the air conditioning, the traffic, the electrical hiss of the preamplifier. Below about −60 dBFS is excellent, and anything above −45 dBFS is going to be audible behind your voice on any recording.

The number matters because of the gap between it and your speaking level. If you speak at −18 dBFS and the room sits at −50 dBFS, you have 32 dB of separation and a clean recording. If the room sits at −35 dBFS, that gap has closed to 17 dB and listeners will hear the room behind every sentence. Closing the gap is usually about moving closer to the microphone, not turning anything up.

Testing consistency, not just presence

The second thing people skip is whether the level stays put. Record thirty seconds while doing what you actually do on a call — turning to look at a second screen, leaning back to think, gesturing. Play it back and listen for your voice rising and falling.

Large swings mean the microphone is too far away, because the inverse square law makes distance changes enormous when you start far out and trivial when you start close. Moving from 60 cm to 90 cm loses more than half your level; moving from 15 cm to 20 cm barely registers. This is why close microphone placement solves problems that look like they need better equipment.

Test in the application, not only in the browser

A browser test proves the hardware, the driver and the operating system are all fine. It cannot prove your meeting application is configured correctly, because every application keeps its own device selection and its own processing chain.

Zoom, Teams, Discord and Meet all have a built-in audio test that records a short clip and plays it back. Use it, because it is the only way to hear what that application's processing does to your voice — and each of them applies different noise suppression and gating.

A five-minute routine before anything important

None of this needs to be repeated daily. What is worth repeating before an interview, a recording session or a presentation is short: confirm the right device is selected, speak at your normal volume and check that peaks land near −12 dBFS, record five seconds and listen on headphones, and be silent for ten seconds to confirm nothing new has started humming in the room.

That routine catches the failures that actually happen — a headset that reconnected and changed the default device, a level that reset after an update, a new noise source — and it takes less time than apologising for bad audio once.

More guides