A free online mic test. Press Start test below and allow the microphone, and the level, waveform and spectrum appear right away along with peak level, noise floor, signal to noise ratio and clipping. Every reading is in dBFS, relative to digital full scale, not room sound pressure.
Press the button above, then choose Allow in the prompt on the left of the browser address bar. If no prompt appears, check the site settings in your browser first. Click here for the detailed Chrome permission guide. Type chrome://settings/content in the address bar to check Chrome site settings, or click here for the operating system permission steps.
Denoise filters steady fan or keyboard noise, Echo cancel suppresses nearby speakers, and Auto gain helps when you move closer and farther from the mic. Leave all three options unchecked when you want to test the microphone as it really sounds.
− · − ch · −
Press Start test, then talk at your usual volume for a few seconds.
Press Start test, then talk at your usual volume for a few seconds.
Record 5 s, play it back, download the WAV or listen live with headphones.
Press Record 5 s when you are ready
Wear headphones before using Listen live, or the speakers will feed back into the mic.
3D needs WebGL. Your browser does not support it, so the 2D view stays on.
Press Start test to see the live waveform here
MicTester has one page and one job: telling you whether your microphone is working and what the signal actually looks like. Most built in checks in other software only say a microphone was detected, but the real problem is usually a low input level, a noisy room or auto gain quietly switched on, which leaves your voice sounding poor.
Every number on this page describes what is actually happening to your voice, and they all come from the same stream the browser hands to the page. Peak and speech level come from time domain samples, the noise floor is estimated from the quietest stretches, SNR is the gap between them, the clip count marks samples at full scale, and the spectrum is a fast Fourier transform of the same data. All of it is there to help you find exactly where the problem is and how to fix it.
The tool is free, needs no account and runs entirely in the browser. It is a quick check before a call, a recording session or a support ticket. If the mic works here but fails in another app, this page at least lets you rule out the microphone itself and narrow things down.
| Web Audio API | Level, waveform, spectrum and sample rate readings come from AudioContext and AnalyserNode as implemented by your browser. |
|---|---|
| Media Capture and Streams | Device selection and the three processing switches are getUserMedia constraints; a browser may ignore a constraint it cannot satisfy. |
| dBFS convention | 0 dBFS is digital full scale. We quote -30 to -6 dBFS for speech peaks and -45 dBFS as a noise floor reference as practical guidance, not a formal standard. |
No sign up and nothing to install. Just open the browser and three simple steps are enough to test your microphone.
Press Start test, then choose Allow in the browser prompt. On a desktop the prompt sits near the address bar; on a phone it appears at the top of the screen.
→ The status changes to Listening and the level bar starts to move with your voice.
Talk for a few seconds from the distance you normally use, or clap once. Watch the level, waveform and spectrum at the same time.
→ Speech peaks should land between -30 and -6 dBFS.
Check the noise floor, SNR and clip count. Move closer, lower the system input volume, or turn off processing, then test again.
→ Zero clips and a noise floor below -45 dBFS are a healthy baseline.
Run this test first, then repeat it inside the app. If both pass, the problem sits in that app rather than in the microphone.
When you speak, your voice passes through the microphone, the sound card, the operating system and the browser before it reaches this page. An online mic test only reads the last step, and every step along the way can change the final result.
Turns sound pressure into an analogue signal, and the capsule and its sensitivity decide how much noise you start with.
Converts the analogue signal into numbers, usually at 48 kHz, and that is where the sample rate shown in the tool comes from.
Input volume and mic boost raise the signal before the browser sees it. A boost also raises the noise floor by the same amount.
Applies denoise, echo cancellation and auto gain when those switches are on, and each one changes the final result.
Reads level, noise floor, clipping and dropouts from the final stream, and it cannot see what happened earlier in the chain.
dBFS is a digital scale where 0 is the largest signal the format can hold, so everyday readings are negative. It says nothing about how loud the room is. To get dB SPL you would need the microphone sensitivity in mV/Pa and a calibrated sound level meter, which is why this mic test reports dBFS only.
Whether sound is arriving, plus level, clipping, noise floor, signal to noise ratio and dropouts.
Room loudness in dB SPL. A page only receives dBFS, has no microphone sensitivity data and has no calibrated reference.
Denoise lowers the noise floor. Switch all three processing options off to measure the raw microphone.
If the level never moves, the browser may not have permission at all. Here is where each platform keeps that switch.
| Platform | Where to allow the microphone | What to check |
|---|---|---|
| Windows 11 | Settings > Privacy & security > Microphone | Turn on Microphone access and Let desktop apps access your microphone |
| macOS | System Settings > Privacy & Security > Microphone | Enable the switch for the browser you are using, then restart the browser |
| iPhone, Safari or Chrome | Settings > Apps > your browser > Microphone | Turn the switch on, then reload this page |
| Android, Chrome | Settings > Apps > your browser > Permissions > Microphone | Set to Allow while using the app, then reload this page |
Work through these before pressing Try again in the tool.
The mic test gives you six core numbers. Each one answers a different question, and read together they tell you whether the problem is level, noise or clipping.
Sound is coming in, but the peaks stay very low. Call apps may add gain on their own, so the other side hears amplified hiss and a thin voice. Try moving closer to the mic or raising the system input volume.
Loud enough for the other person to hear you clearly, with room left for a sudden loud moment. This is the comfortable range, so you can keep these settings and start using the mic.
You have reached the top of the range, and this is where the clip count starts to climb. Lower the input volume or move the mic a bit further away, because once audio clips, no amount of turning the volume down later will bring it back.
| Reading | What it measures | Healthy range | If it is off |
|---|---|---|---|
| Peak level | Loudest sample in the last 1.4 seconds | -30 to -6 dBFS | Lower input volume or step back if it reaches 0 |
| Speech level | Average power of the loudest recent frames | -30 to -12 dBFS | Raise input volume or move closer to the mic |
| Noise floor | Quietest stretch while you pause | Below -45 dBFS | Remove the noise source, lower mic boost, then consider denoise |
| SNR | Gap between speech level and noise floor | Above 20 dB | Improve the room or the mic placement before adding processing |
| Clips | Runs of samples pinned at full scale | 0 | Back off gain and check mic boost and the system level |
| Input latency | Delay from capture to this page | Under 20 ms | Use a wired connection and close heavy tabs |
If you want to understand how the test works, these 12 terms are the ones that come up most often in a mic test.
A digital scale where 0 is the largest value a sample can hold. Everything below it is negative, which is why a healthy mic reads around -12 dBFS and not 12.
The single loudest sample in a window. It decides whether you clip and reacts faster than any average.
The average power of the signal over a short window. It tracks perceived loudness better than peak, and it is what speech level uses here.
The level the meter falls to when nobody speaks. Fans, mains hum and the self noise of the capsule all sit here.
Speech level minus noise floor. 20 dB is workable, 30 dB and above sounds clean on a call.
Samples pinned at full scale. The waveform is flattened, and no later volume change can bring back the lost detail.
The gap between your loudest peak and 0 dBFS. Keeping 6 dB absorbs a sudden laugh or a dropped word.
The delay between sound reaching the capsule and the page receiving it. Above roughly 20 ms, live monitoring starts to feel disconnected.
How often the sound card measures the signal, usually 44.1 or 48 kHz. Bluetooth headsets drop to 16 kHz in call mode.
Processing that raises quiet parts and pulls back loud ones. It hides the real level and lifts the noise floor, so switch it off when comparing mics.
How a microphone favours some frequencies over others. It explains why one mic sounds warm and another thin, and a level test cannot show it.
Routing your own input back to headphones so you hear yourself while speaking. Always use headphones, or the speakers will feed back.
Your microphone is how people hear you online. If it stops working, every call, stream or class falls apart, so it is worth running a mic test before any of the activities below.
Check the level before a client call instead of discovering a muted headset two minutes in.
Watch peak level and clips to keep your volume in a comfortable range for the call.Confirm the mic you meant to use is the one the browser picked, especially after plugging in a new interface.
Check the device name and the noise floor so you do not pick the wrong mic or let other sounds in.Compare two mics on identical settings: same distance, same input volume, processing off.
Watch speech level and SNR to get the best voice quality you can.Test in the browser, then in the game, to separate a hardware fault from a settings problem.
Look for gaps in the waveform, so your team can still hear you at the moment that matters.Find out whether the laptop mic picks up the whole room before a lesson starts.
Check the noise floor and SNR so the class does not hear background noise when you answer a question.Attach a reading instead of describing the problem. A clip count and a noise floor value are hard to argue with.
All six readings help: give the support team real numbers and the problem gets solved faster.When a mic sounds bad, it is usually about placement and gain, not a broken microphone. These six habits do more than any software setting.
Speak across the capsule rather than straight into it, about a hand width away, which reduces plosives and keeps the level steady.
Set the system input so normal speech peaks near -12 dBFS. Fix the level there first; software gain only amplifies the noise you already have.
Some words push a puff of air at the capsule when you speak. A pop filter or a slight angle gets rid of the thump and makes it sound better.
A bare wall behind the mic adds a hollow tail to your voice, and a curtain, a rug or a bookshelf in the room does more than any denoise switch.
Condenser mics do not like damp. Storing one somewhere dry when it is not in use will make it last longer, and if you bring it in from the cold, let it reach room temperature before powering it on.
Brush the grille with a dry soft brush, never spray cleaner toward the capsule, and do not blow dust off with compressed air, because that airflow can tear a thin diaphragm.
How to read the numbers, where permissions live, and what to do when you move to another app or a phone. If something is already broken, start with the answers on levels, clipping and dropouts.
Start mic test ↗No. The stream is read by an AudioContext inside this tab, so it never leaves the page and is never written to storage. Reload or close the tab and the recording is gone. The analytics scripts only send visit data and cannot read the microphone.
dBFS counts down from digital full scale: 0 is the largest signal a sample can hold, so normal readings are negative, and -6 dBFS is roughly half the full amplitude. It is not a room measurement. Sound pressure in dB SPL needs microphone sensitivity and a calibrated meter, and a browser has neither, so this tool only reports dBFS.
Speech peaks between -30 and -6 dBFS: loud enough for a call to sound clear, with headroom left before clipping. Peaks that never reach -30 dBFS mean the mic is too quiet, so sit closer and raise the system input volume to 70 or above. Try both before switching on mic boost, which lifts the noise floor along with your voice.
The signal has reached digital full scale and everything above it is flattened, which is what makes a clipped recording sound torn. The usual causes are sitting too close, system input volume set too high, or mic boost switched on. Move back, lower the input volume until peaks sit near -6 dBFS, and turn boost off. A clipped passage cannot be repaired afterwards, so keep that headroom while you record.
In a quiet room most microphones sit below -60 dBFS. Above -45 dBFS the hiss or fan noise becomes audible to the listener. Turn off auto gain and mic boost before judging this, because both raise the floor along with your voice.
That is expected. Noise suppression reduces steady background sound, so the floor you saw was a processed result, not the microphone noise itself. Switch off Denoise, Echo cancel and Auto gain together to measure the raw signal.
Use Record 5 s and play the clip back. That single loop exercises the input and the output, so you can confirm both work. Wear headphones or keep the volume low, otherwise the speakers feed back into the mic.
Dropouts are usually a connection problem rather than a broken capsule. Bluetooth is the common case: a low battery, distance, or interference from Wi-Fi and USB 3.0 ports all cut packets, and a wired headset is the quickest way to confirm it. On a wired mic, check that the USB or 3.5 mm plug seats fully and that the socket is not loose; front panel jacks on desktops often suffer from poor internal cabling. A busy system can starve the audio thread too, so close heavy apps and test again.
Check in this order. First, a hardware mute switch, which many headsets put on the cable. Second, the device selector in the tool, in case it picked a webcam or a virtual device. Third, the browser permission for this site; if it is blocked, the level stays at the floor. If all three are fine, the plug is the next suspect: a single 3.5 mm 4-pole plug needs a 4-pole socket or a splitter, and a desktop 3-pole mic jack leaves the mic contact open. With two plugs, the mic goes into the pink jack. Bluetooth headsets switch to a narrow call mode once the mic is active, so they work but sound dull.
Yes. Open this page in Chrome or Safari on iOS or Android and allow the microphone prompt. The steps are the same as on a desktop. Keep in mind that mobile systems apply their own processing, so readings are a check of the whole chain rather than the bare capsule.
Put on headphones, start the test, and press Listen live to route the input straight to the output. You can also record five seconds and play it back, which avoids feedback entirely. A short latency is normal: the signal passes through the sound card, the browser and your output, which adds roughly 10 to 30 ms. A tone that keeps rising is different, and means the speakers are feeding back into the mic, so switch to headphones.
Start the test once so the browser reveals device names, then open the Microphone selector under Device and processing. Pick the input you want and the stream restarts. USB headsets and webcams appear as their own entries, separate from the built in microphone.
Automatic gain control continuously raises quiet parts and pulls back loud ones. That lifts the noise floor, hides the real peak, and can make two microphones look equally loud. For a comparison, turn Auto gain off and keep the distance and system input volume the same.
That result narrows the problem down. The microphone, the system input and the browser are all working, so check the other app input device selection, exclusive mode on Windows, and its own permission settings.
Open this page in Edge or Chrome, press Start test and allow access. If the level stays flat, go to Settings > Privacy and security > Microphone and turn on both switches, then open Settings > System > Sound > Input and watch the bar while you speak. The six checks in the Windows section above cover the rest.
Test here first, so you know the microphone and the system input work. Then open the app audio settings, pick the same input device and run its built in test: Test Mic in Zoom, Make a test call in Teams, Mic Test under Voice and Video in Discord. If the app shows no input while this page does, the fault is in that app settings or its permission.
Level and tone are different problems. A muffled mic is usually off axis, behind a hand or a laptop lid, or running in a narrow call mode, which is common over Bluetooth. A thin sound often comes from a small capsule or from heavy denoise. If peaks and SNR read healthy, the cause is placement or tone rather than signal.
One click, one permission prompt, and you will know what to fix.
