Skip to content

Mic Test

Check that your microphone works: live level meter, waveform, spectrum, device details and a test recording, all local.

  • Free
  • No sign-up
  • Runs locally

Runs in your browser; what you enter is not uploaded to our servers.

Updated

What this mic test does

This mic test tells you in a few seconds whether your microphone works and how good the signal is. Press Start mic test, allow access, and speak. A live meter shows the level in dBFS, a waveform and a frequency spectrum show what the microphone hears, and a status line answers the real question: is my mic working, too quiet, too loud, or sending pure silence?

It also shows which device the browser opened, its sample rate and channel count, and whether echo cancellation, noise suppression and auto gain are active. You can switch microphones, turn each filter on or off, and record a 5, 10 or 30-second clip to hear exactly what other people hear in a call. The audio is analyzed in your browser and never leaves your device.

How to test your microphone

  1. Press Start mic test. The browser asks for permission. Choose Allow. Nothing is requested until you click.
  2. Speak at your normal distance. Count to ten or read a sentence aloud. The status line changes from Listening to a verdict within a second or two.
  3. Pick the right device in the Microphone list if the wrong one opened. The list fills in once access is allowed, and the test restarts on the new device.
  4. Compare the filters. Echo cancellation, noise suppression and auto gain control are on by default, as in most video calls. Switch them off to hear the raw microphone.
  5. Record a clip and play it back. Download it if you want to keep it, or press Delete.
  6. Press Stop mic test when you are done.

If the browser does not ask, or you see an error, the message names the cause and the setting to change.

Reading the meter: dBFS, RMS and peak

Digital audio is a stream of numbers between −1 and +1. The meter converts them to decibels relative to full scale (dBFS), where 0 dBFS is the largest value the signal can hold:

level in dBFS = 20 × log10(amplitude)

amplitude 1.0      ->    0 dBFS   (the ceiling)
amplitude 0.5      ->   −6.02 dBFS
amplitude 0.1      ->  −20 dBFS
amplitude 0.01     ->  −40 dBFS

Halving the amplitude costs about 6 dB; a tenth costs 20 dB. Readings are negative because nothing goes above the ceiling.

The meter shows two values. Peak follows the largest single sample in each window of 2,048 samples (about 43 ms at 48,000 Hz). RMS (root mean square) is the average energy and is closer to how loud the sound feels. Both rise instantly and fall back gradually, RMS at 30 dB per second, so they stay readable. For a pure tone that reaches full scale, the peak is 0 dBFS and the RMS is −3.01 dBFS, because the RMS of a sine wave is its peak divided by √2. Some recording software adds 3.01 dB so that tone reads 0 dB RMS; this tool does not.

The thin line is a peak hold: it marks the highest recent peak for 1.5 seconds, then falls at 20 dB per second. A sample at 99% of full scale or more (−0.09 dBFS) counts as clipping, and the CLIP light stays on for 1.5 seconds so you notice short overloads.

What the meter shows (speech, last 3 s)VerdictWhat to do
Every sample exactly 0Pure silenceSomething mutes the input; see troubleshooting
Peaks below −50 dBFSVery little soundSpeak, or raise the input volume
Peaks from −50 to −24 dBFSWorking, but quietMove closer or raise the input volume
Peaks from −24 to −3 dBFSWorking, healthy levelNothing
Peaks from −3 to 0 dBFSWorking, loudLower the input volume a little
Any sample at 99% of full scale or moreClippingLower the volume or move back

These bands are practical guidance for speech, not a standard. For recording, peaks around −12 dBFS leave room for a sudden laugh.

The noise floor tile is the quietest 10% of RMS readings over the last 10 seconds. It is how loud the room and the microphone are when you are not speaking. The bigger the gap between your speech and that number, the cleaner you sound.

Waveform and spectrum: what they reveal

The waveform draws the latest window of samples. Speech shows as bursts around the center line. A flat line while you talk means no signal reaches the browser. Tops cut off flat at the dashed lines mean the signal is clipping.

The spectrum splits the sound into frequencies on a logarithmic axis from 40 Hz to 16 kHz. It uses an 8,192-point analysis, so at 48,000 Hz each bin is 48,000 ÷ 8,192 = 5.86 Hz wide. Three patterns are worth knowing:

  • A spike at 50 or 60 Hz (or 100, 120, 150 or 180 Hz) is usually mains hum. Mains power runs at 50 Hz in Europe, most of Asia, Africa and Australia and at 60 Hz in North America and parts of South America and Asia. The tool names the hum when the loudest tone sits within 4 Hz of these frequencies. Ground loops, unshielded cables and power adapters near the mic are the usual sources.
  • Nothing above 4 kHz or 8 kHz means the microphone samples at 8,000 or 16,000 Hz, since a digital signal can only hold frequencies up to half its sample rate. Most Bluetooth headsets switch to this hands-free mode whenever their microphone is in use.
  • A broad bump that never goes away is steady noise such as a fan or air conditioner. Noise suppression is designed for exactly this.

Troubleshooting: when the mic is not working

Windows 11. Open Settings › Privacy & security › Microphone and turn on Microphone access and Let desktop apps access your microphone. Browsers are desktop apps, so the second switch matters. Then open Settings › System › Sound, pick your device under Input and check that its volume is not at zero. On Windows 10 the privacy page is Settings › Privacy › Microphone. If the error says the device could not be started, another app may hold it exclusively: in the classic Sound control panel, open Recording › your microphone › Properties › Advanced and clear Allow applications to take exclusive control of this device.

macOS. In System Settings › Privacy & Security › Microphone, turn on your browser. In System Settings › Sound › Input, select the microphone and watch its input level while you speak.

Chrome and Edge. Click the icon left of the address bar and set Microphone to Allow, then reload. The full list of blocked sites and the default input device are at chrome://settings/content/microphone or edge://settings/content/microphone.

Firefox. Click the crossed-out microphone in the address bar to remove a block, or open Settings › Privacy & Security › Permissions › Microphone. Firefox also lets you pick the device in its permission prompt.

Safari, iPhone and iPad. On a Mac, open Safari › Settings › Websites › Microphone. On iOS, use the page-settings button in the address bar, or Settings › Safari › Microphone (Settings › Apps › Safari on iOS 18). Chrome and Firefox on iOS have their own Microphone switch in the iOS Settings app.

Android. In Chrome, tap the icon left of the address and allow Microphone, and check Settings › Apps › Chrome › Permissions. On Android 12 and later, the Microphone access tile in Quick Settings blocks every app when it is off.

Hardware. Look for a mute switch on the headset cable or ear cup, a mic-mute key on the laptop keyboard, and a mute button or gain knob on USB microphones. A phone-style headset with one 4-pole plug needs a combined headset jack; a computer with separate pink and green jacks needs a splitter. If the wrong device keeps opening, such as a webcam microphone or a virtual device from meeting software, choose the right one here and set it as the default in your system sound settings.

Practical checks before it matters

  • Before an interview or exam call: run the test, record 10 seconds with the filters on and listen with headphones. Do the same in the room and position you will use.
  • Podcasts and voice-overs: turn all three filters off, aim for speech peaks near −12 dBFS and check that the noise floor sits well below −60 dBFS.
  • The test works, the call does not: then the calling app has its own input selection, mute button or permission. The keyboard and mouse can be checked the same way with the keyboard tester and the mouse test.

Privacy and limits

The microphone stream goes only to analyzers running in this tab. No audio, level or device name is sent anywhere. Recordings are created only when you press Record, held in memory as a local file and removed when you delete them or close the tab. The chosen microphone and the three filter settings are saved in this browser, so the next test starts with them. Your system shows its own indicator while the test runs (an orange dot on Mac and iPhone, a green one on Android), and it disappears when you press Stop.

Limits: the readings describe the digital signal the browser receives, after any processing by the system, the driver or the headset, so they cannot measure sound pressure in dB SPL. Browsers may ignore a requested filter setting; the Detected input panel shows what is actually active, where reported.

Frequently asked questions

The meter moves here, so why can people in my call not hear me?

Then the microphone and the browser work, and the problem is in the calling app. Check the input device chosen inside the app (Zoom, Teams, Discord and others keep their own setting), make sure you are not muted there or using push-to-talk, and check that the app itself has microphone permission in your system privacy settings. Desktop apps and websites are allowed separately.

Why does the test say my microphone sends pure silence?

A working microphone always picks up a little noise, so a signal of exact zeros means something is switching it off. The usual causes are a mute key or switch on the laptop or headset, the input muted or set to zero in the system sound settings, a privacy setting that blocks the browser, or a device selected that has nothing connected, such as an empty jack.

What is a good microphone level in dBFS?

For speech, this test treats peaks between −24 and −3 dBFS as healthy. 0 dBFS is the loudest value a digital signal can hold, and anything that reaches it clips and distorts, so peaks around −12 dBFS leave room for sudden laughs or loud words. If normal speech stays below −24 dBFS, move closer or raise the input volume.

How do I test a headset or Bluetooth microphone?

Connect the headset, press Start and choose it in the Microphone list. Most Bluetooth headsets switch to a hands-free mode when their microphone is used, which samples at 8 or 16 kHz, so the spectrum stops at about 4 or 8 kHz and your voice sounds like a phone call. That is normal for Bluetooth, not a fault.

Does this mic test record or upload my voice?

No upload happens at any point. The level, waveform and spectrum are calculated from the live signal inside your browser and are not saved. A recording is made only when you press a Record button; it is kept in this tab's memory, plays back from there and disappears when you delete it or close the tab, unless you choose Download.

Why does my voice sound different in calls than in the test recording?

Calls usually process the signal: echo cancellation removes sound from your speakers, noise suppression filters steady background noise and auto gain control raises or lowers the volume. Those filters can make a voice sound thinner or make the level pump. Turn the three switches off and on here, record a clip each way and compare.

Why does my browser not ask for microphone permission?

Most often the site was blocked earlier, and the browser remembers it without asking again. Reset the permission from the icon to the left of the address bar. Browsers also refuse silently on pages without https, inside some in-app browsers in social and chat apps, and on managed work or school devices where an administrator has turned the microphone off.

What sample rate should my microphone use?

48,000 Hz and 44,100 Hz are the usual rates for USB microphones and built-in laptop mics, and both cover the full range of human hearing. The Detected input panel shows what the browser opened. Readings of 16,000 or 8,000 Hz usually mean a Bluetooth headset in call mode, which is enough for speech but loses the higher frequencies.

Published · Updated

Found a bug or have an idea? Tell us