Skip to main content

Microphone test online

Verify your microphone with real-time level, waveform, and spectrum analysis.

Free & unlimitedWorks offline

Microphone access needed

This tool reads your microphone to measure level, waveform, and spectrum. Audio is processed locally in your browser — nothing is uploaded.

All processing happens in your browser. No data is sent to any server.

This microphone test opens an audio-only stream with getUserMedia and routes it into a Web Audio AnalyserNode with a 2048-point FFT. From the time-domain buffer it computes RMS and converts it to decibels on a 0 to -60 dBFS scale, holds the peak for a second and then lets it decay, and raises a clipping warning the moment level passes -1 dBFS. The same buffer draws a live oscilloscope, while the frequency bins draw a spectrum so you can see hum, rumble and sibilance rather than guess at them. Noise suppression, echo cancellation and auto gain control start switched off, so what you see is the raw capture.

Key facts about Microphone test online

Key facts about Microphone test online
Capture APInavigator.mediaDevices.getUserMedia with audio constraints and an optional deviceId
Analysis nodeWeb Audio AnalyserNode, fftSize 2048, smoothingTimeConstant 0.8
Level calculationRMS of the 2048-sample time-domain buffer, converted with 20 x log10
Meter rangeclamped to 0 dBFS at the top and -60 dBFS at the floor
Peak holdheld for one second, then decays 0.5 dB every 50 ms back toward the floor
Clipping thresholdraised above -1 dBFS, where digital headroom is effectively gone
OscilloscopegetByteTimeDomainData, redrawn every animation frame
SpectrumgetByteFrequencyData across 1024 bins, up to half the sample rate
Processing togglesnoise suppression, echo cancellation and auto gain control, all off by default
Device infotrack label, AudioContext sample rate, channel count and baseLatency in milliseconds
Recording formatMediaRecorder using audio/webm with Opus when the browser supports it
Exporta JSON report with the device info, the peak level reached and the processing settings used

What happens to your file

Audio is analysed in this tab and nowhere else. The stream from getUserMedia is connected to an AnalyserNode inside a local AudioContext, the numbers it produces are drawn straight onto two canvases, and no sample is ever sent over the network. A test recording is assembled by MediaRecorder into a WebM Blob held in memory; play it back, download it, or navigate away and it is gone. Stopping the mic stops every track, closes the AudioContext and releases the device, which is what turns off the recording indicator in your operating system. Nothing is written to storage, so there is no history of what you said.

About this tool

  1. 1

    Grant microphone access

    Accept the browser prompt. If you blocked the microphone here before, the prompt will not return — clear the site permission from the padlock in the address bar and reload.

  2. 2

    Select the right input

    Once permission is granted, the device list shows real names. Laptops usually have both a built-in array and whatever headset is plugged in; pick the one you actually want to test.

  3. 3

    Speak at your normal distance

    Talk at the volume and distance you would use in a call. Aim for peaks around -12 to -6 dBFS: loud enough to sit well above the noise floor, quiet enough to leave headroom.

  4. 4

    Check the noise floor in silence

    Stop talking and watch the meter settle. A quiet room with a decent mic sits near the -50 to -60 dBFS floor. A floor that hovers around -30 means hum, fan noise or a gain setting that is far too high.

  5. 5

    Read the spectrum, not just the meter

    A strong bar at the far left is low rumble from a desk or an air conditioner. A narrow spike at 50 or 60 Hz and its multiples is mains hum. A hot region between 5 and 8 kHz is sibilance.

  6. 6

    Toggle the processing on and off

    Noise suppression, echo cancellation and auto gain control start off so you hear the raw capture. Turn each on and watch the spectrum and meter change — that is exactly what a call app is doing to your voice.

  7. 7

    Record and listen back

    A short recording played through your headphones is the only way to hear artefacts a meter cannot show: crackle from a failing cable, USB whine, room echo and plosives.

  8. 8

    Export the report if you are filing a ticket

    The JSON summary records the device name, sample rate, channel count, base latency and the peak level reached, which is more useful to a support desk than a description.

Specs & compatibility
Browser supportChrome, Edge, Firefox, Safari 11+ and Opera; Safari needs the webkitAudioContext fallback the page includes
Secure contextmicrophone access requires HTTPS, which this page uses
Sample rateset by the AudioContext, normally 44,100 or 48,000 Hz — the page does not force a rate
Spectrum ceilinghalf the sample rate, so roughly 22 kHz or 24 kHz
Device labelsblank until permission is granted, by browser design
Safari and the togglesmay ignore noiseSuppression and autoGainControl constraints and apply its own processing
Recording playbackWebM Opus plays in Chrome, Edge and Firefox; Safari support for the container is limited
Account or installnone — no signup, no driver, no app
  • Wear headphones before enabling the mic. Open speakers plus a live input is how feedback loops start.
  • Clipping is a hard ceiling, not a loudness rating: once the meter hits it, that part of the waveform is gone and no software can restore it. Lower input gain in the OS sound settings rather than moving further away.
  • A mains hum spike at 50 Hz in Europe or 60 Hz in North America is usually a ground loop; try a different USB port, remove hubs, or power a laptop from battery to test.
  • Auto gain control makes a quiet mic look fine on the meter by amplifying the room with it — test with it off first, then decide.
  • Echo cancellation is designed around speakers and a mic in the same room. Leave it on for calls and off when recording anything you intend to keep.
  • If the level meter never moves, check the operating system input device and its volume slider; the browser can only read what the OS routes to it.
  • Base latency below about 10 ms means the audio stack is behaving. Much larger figures often point at a Bluetooth headset or a shared driver mode.
  • Bluetooth headsets drop to a narrowband profile when the microphone is in use, which caps the spectrum near 8 kHz — that is the protocol, not the microphone.
  • Live dB meter with one-second peak hold and decay
  • Clipping warning above -1 dBFS
  • Oscilloscope drawn from the raw time-domain buffer
  • Frequency spectrum from a 2048-point FFT
  • Noise suppression, echo cancellation and auto gain toggles
  • Device selector with sample rate, channels and base latency
  • Record, play back and download a WebM test clip
  • JSON report export
  • Checking a headset works before a meeting, an interview or a stream.
  • Finding the source of a hum or buzz by reading it off the spectrum instead of guessing.
  • Setting input gain so peaks land near -12 dBFS with no clipping.
  • Comparing a USB microphone against a laptop's built-in array on the same scale.
  • Testing whether a browser's noise suppression is what makes your voice sound thin on calls.
  • Documenting a faulty microphone with a recorded clip and a device report for support.
They are dBFS, decibels relative to digital full scale, not loudness in the room. Zero is the maximum a digital signal can represent and everything below it is negative; the meter floors at -60. A comfortable speaking level peaks around -12 to -6 dBFS, which leaves headroom for a laugh or a raised voice without clipping. The number is computed from the RMS of a 2048-sample window, so it tracks average level rather than instantaneous spikes, which is why the separate peak indicator matters.
Clipping is set by input gain, not by perceived loudness. If the operating system's input level or a microphone's own gain knob is set high, even normal speech can push the signal past -1 dBFS, where the top of the waveform is flattened and the detail is discarded. Lower the input level in your sound settings until peaks sit around -12 to -6 dBFS. Moving further from the microphone also works but changes the tone; reducing gain is usually the better fix.
Start with everything off, which is how the tool opens. That shows what your microphone and room actually sound like, including the noise floor, hum and echo you might otherwise blame on the call software. Then switch the toggles on one at a time to see what each removes. Noise suppression and auto gain control both make the meter look better while changing your voice, so for judging hardware they should be off; for day-to-day calls they usually help.
No. The audio is captured by MediaRecorder inside this tab and collected into a WebM Blob held in memory. Playback reads that Blob through an audio element in the same tab, and the download button hands it to your browser as a local file. No sample is transmitted, nothing is written to local storage or cookies, and there is no server component to this page at all. Navigating away or closing the tab discards the clip permanently.
The most common reason is a Bluetooth headset. When a Bluetooth device's microphone is active, the connection switches to a narrowband or wideband voice profile whose bandwidth stops well below the music profile's, so nothing above roughly 8 kHz exists to display. Aggressive noise suppression can also strip high frequencies. Test the same headset over its USB dongle or cable to compare, and remember that the spectrum can never show more than half the AudioContext sample rate.
Work outward from the browser. Check that the permission was granted and that the correct device is selected in the tool's own list. Then check the operating system: the input device, its level slider, and whether the microphone is muted at hardware level, which many headsets do with an inline switch or a boom that mutes when raised. On Windows also check the app-level microphone privacy setting for your browser. Finally, close any other application that may be holding the device exclusively.

Related tools

View all

Part of Hardware testers that run in your browser

Related categories2

Updated

We use anonymous analytics to improve ToolChamp. No personal data is stored or sold. Privacy Policy