Microphone Test

How do I test my microphone in the browser?

Open AnyTool Microphone Test, press Start test and allow the microphone when your browser asks. Speak, tap or clap and the animated level meter and live waveform react to your voice — that movement proves the mic is working. The tool reads the active audio track and the Web Audio engine to show the microphone’s name, sample rate, channel count and reported latency. If you have more than one mic, pick another from the dropdown and the test restarts on it. Press Record to capture a short clip, then play it straight back to hear how you actually sound, and download it if you like. Everything runs locally — your microphone audio is analysed only on your device and is never uploaded; recordings stay in your browser unless you choose to download them, and the mic stops the moment you press Stop test or leave the page.

Animated level (VU) meter + live waveform that react to your voice — proves the mic works
Shows the real mic name, sample rate, channel count and latency (track.getSettings + AudioContext)
Switch between multiple microphones with the device selector (enumerateDevices)
Record a short clip and play it back locally to hear yourself — download optional
100% client-side: your audio is never uploaded and the mic isn’t routed to your speakers

What is

A microphone test confirms that your mic works and shows what it is capturing: a live level meter and waveform that move when you speak, plus the device’s name, sample rate and channel count. A browser-based test uses the MediaDevices getUserMedia API to open the mic with your permission, feeds it into a Web Audio AnalyserNode to visualise the signal, and can record a short clip with the MediaRecorder API for instant playback — all on your device, with nothing uploaded.

Generators

Related terms

getUserMediaWeb Audio APIAnalyserNodeMediaRecorderenumerateDevicesRMS / level meterWaveformSample rate

Frequently Asked Questions

Everything happens locally. The browser opens the microphone with getUserMedia only after you grant permission, and the stream feeds a Web Audio AnalyserNode purely to draw the level meter and waveform — there is no server, no account and nothing transmitted. The mic is deliberately not connected to your speakers, so there is no feedback. If you press Record, the MediaRecorder API captures a short clip as an in-memory blob that plays back in an audio element on your device; it is written to disk only if you press Download. The microphone track is stopped and the audio engine closed the moment you press Stop test, leave the page or close the tab, which turns the mic indicator off.

No. Your audio is analysed only on your device, live in your browser, and is never uploaded. Any clip you record stays in your browser as a temporary blob and is only saved if you download it; the mic stops the instant you press Stop test or leave the page.

A flat level meter while you speak almost always comes down to a few things: the wrong microphone is chosen in the dropdown, a hardware mute switch or your OS mixer has the mic muted, the browser hasn’t been granted permission (click the mic icon in the address bar to allow it), or another app is holding the device — close Zoom, Teams, Meet, OBS or another browser tab. Microphone access also needs a secure (https) context. Once the right, unmuted mic is selected and permission is granted, speaking, tapping or clapping near it should make the bar jump and the waveform wiggle.

Usually the wrong mic is selected, the mic is muted, permission is missing, or another app is using it. Pick the right device in the dropdown, unmute it, allow the mic for the site, and close Zoom, Teams, Meet or OBS.

After starting the test, the Record button uses the MediaRecorder API on the live microphone stream to collect a short clip into an in-memory blob. When you stop, the clip appears with an audio player so you can immediately play it back and judge clarity, volume and background noise — the single best way to confirm the mic sounds right. The recording is held only in your browser; you can download it as an audio file, discard it, or keep several clips to compare. Nothing is uploaded at any point.

Yes. Press Record to capture a short clip with the MediaRecorder API, then play it back in the built-in audio player to hear how you sound — entirely local. You can download the clip or discard it, and keep a few at once.

When the test starts, the tool takes the live audio track from the MediaStream and calls track.getSettings(), which returns details such as the channel count, deviceId and reported latency, plus track.label for the human-readable microphone name. The Web Audio AudioContext provides the sample rate it is running at. These are the values the browser and device negotiated for this session rather than the hardware’s theoretical maximum, so they describe how the mic is running right now. The on-screen dBFS figure is a relative loudness reading derived from the signal, not a calibrated SPL measurement.

It reads the active audio track’s getSettings() and the Web Audio AudioContext, which report the device name, sample rate, channel count and reported latency for this session.

Microphone Test

A free online microphone test with an animated level (VU) meter and live waveform that react to your voice, real device info (name, sample rate, channels, latency), multi-mic switching, and record-and-play-back so you can hear yourself — built on the native getUserMedia, Web Audio and MediaRecorder APIs, entirely in your browser with nothing uploaded.

Features

Animated level meter + live waveform with Start / Stop and a clear permission/empty state
Real device info: microphone name, sample rate, channel count and reported latency
Multi-microphone selector (enumerateDevices) — switching restarts on the chosen device
Record a short clip and play it back locally to hear yourself; download optional
100% client-side — your audio is never uploaded and recordings stay in your browser

PriceFree

CategoryGenerator

PlatformAll (browser-based, requires microphone permission and a secure HTTPS context)

PrivacyThe microphone audio is analysed only on your device via getUserMedia; nothing is uploaded, the mic is not routed to your speakers, and recordings stay in your browser unless you download them

Input: Live microphone input (your mic, with permission)

Output: On-screen live level meter, waveform and device info, Recorded audio clip (.webm/.ogg) played back and downloaded locally

Detailed Explanation

Methodology

How the Microphone Test Opens and Reads the Mic

AnyTool Microphone Test runs entirely on the native MediaDevices and Web Audio APIs — navigator.mediaDevices.getUserMedia, enumerateDevices, AudioContext and AnalyserNode — with no library, CDN or upload. Pressing Start test checks window.isSecureContext and calls getUserMedia({ audio }), then wires the MediaStream through a MediaStreamAudioSourceNode into an AnalyserNode. Crucially the source is never connected to ctx.destination, so the mic is not routed to the speakers and there is no feedback. A requestAnimationFrame loop reads getFloatTimeDomainData each frame to compute the RMS (loudness) and a slow-decaying peak hold for an animated level meter with a dBFS readout, and draws the live time-domain waveform on a canvas. The active audio track’s getSettings() and the AudioContext report the device name (track.label), sample rate, channel count and reported latency. enumerateDevices() lists every audioinput — labels populate once permission is granted — to fill a microphone selector; choosing another mic stops the current tracks and restarts getUserMedia with { deviceId: { exact } }. A track ‘ended’ listener catches a mic unplugged or grabbed by another app mid-session.

getUserMedia({ audio }) → MediaStreamSource → AnalyserNode, NOT connected to the speakers (no feedback)
requestAnimationFrame reads getFloatTimeDomainData → RMS + peak-hold level meter and a live waveform
track.getSettings() + AudioContext → mic name, sample rate, channel count and reported latency
enumerateDevices() populates a mic selector; switching restarts on the chosen deviceId
Secure-context check + a track ‘ended’ listener for unplugged / in-use microphones

How It Works

Record, Play Back and the Level Meter

The single clearest proof a mic works is recording a short clip and hearing yourself, so the tool wraps the MediaRecorder API around the live stream: ondataavailable collects chunks and onstop assembles them into a Blob, which becomes an object URL played in an <audio> element for instant local playback. Clips can be downloaded as audio files (.webm or .ogg depending on the browser), discarded, or kept several at a time to compare; every object URL is revoked when a clip is removed or the page unloads. The animated level meter maps the RMS into a coloured bar with a thin peak-hold marker and an approximate dBFS figure, and surfaces clear states — NO SIGNAL when the bar stays flat, GOOD in the comfortable range, and TOO LOUD near clipping — with a prompt to speak, tap or clap and watch the bar move. The same Start/Stop and Record controls repeat in a sticky mobile action bar.

MediaRecorder → chunks → Blob → object URL → <audio> for instant local playback
Download a clip (.webm/.ogg), discard it, or keep several to compare
Level meter shows RMS, a peak-hold marker and an approximate dBFS readout
Clear NO SIGNAL / GOOD / TOO LOUD states plus a “speak and watch the bar” hint
Object URLs are revoked on discard, clear-all and unmount

Limitations

Privacy and Honest Limits

Because the microphone is opened and analysed entirely in the browser, your audio is processed only on your device and is never uploaded — there is no server, no account and nothing transmitted, and the mic is not routed to your speakers. Any recording is an in-memory blob that stays in your browser and is written to disk only if you choose to download it. The microphone track is stopped, the AudioContext closed, any recorder stopped and object URLs revoked the instant you press Stop test, leave the page or unmount. Honestly, the sample rate, channel count and latency shown are what the browser and device negotiated for this session rather than the hardware’s maximum, and the dBFS figure is a relative loudness reading, not a calibrated SPL measurement. A flat level meter usually means the wrong mic is selected, the mic is muted, permission is missing, or another app (Zoom, Teams, Meet, OBS or another tab) holds the device. Access requires permission and a secure (https) context, and recording relies on the MediaRecorder API, which a few older browsers lack.

Microphone Test: in-browser (AnyTool) vs typical online testers
Capability	AnyTool	Typical online testers
Mic access	Native getUserMedia, secure-context checked	getUserMedia, varies
Level meter	Animated RMS + peak-hold + dBFS readout	Often a basic bar
Waveform	Live Web Audio time-domain canvas	Sometimes none
Device info	Name, sample rate, channels, latency	Often name only
Multi-mic	enumerateDevices selector, live switching	Sometimes single mic
Record + playback	MediaRecorder → local playback + download	Sometimes none
Privacy	Analysed locally, never uploaded; no feedback to speakers	May stream or record remotely

AnyTool analyses your microphone locally with the native getUserMedia, Web Audio and MediaRecorder APIs; your audio is never uploaded, the mic is not routed to your speakers, and recordings stay in your browser unless you download them.

Cite This Page

Title: Microphone Test (free online mic test with an animated level meter and live waveform, real device info — name, sample rate, channels and latency — multi-mic switching, and record-and-play-back so you can hear yourself, native getUserMedia + Web Audio + MediaRecorder, 100% client-side)
Author: AnyTool
Published: January 1, 2026
Modified: February 8, 2026
Section: Generators
URL: https://www.anytool.tech/microphone-test

Formatted Citation

AnyTool. "Microphone Test (free online mic test with an animated level meter and live waveform, real device info — name, sample rate, channels and latency — multi-mic switching, and record-and-play-back so you can hear yourself, native getUserMedia + Web Audio + MediaRecorder, 100% client-side)." AnyTool, January 1, 2026. https://www.anytool.tech/microphone-test

A privacy-first, 100% client-side microphone test built on the native MediaDevices and Web Audio APIs — navigator.mediaDevices.getUserMedia, enumerateDevices, AudioContext, AnalyserNode and MediaRecorder — with no library, CDN or upload. Pressing Start test checks window.isSecureContext and calls getUserMedia({ audio }); the returned MediaStream is wired through a MediaStreamAudioSourceNode into an AnalyserNode that is deliberately not connected to ctx.destination, so the mic is never routed to the speakers and there is no feedback. A requestAnimationFrame loop reads getFloatTimeDomainData each frame to compute the RMS loudness and a slow-decaying peak hold, driving an animated level (VU) meter with a peak-hold marker and an approximate dBFS readout, and draws the live time-domain waveform on a canvas — speak, tap or clap and the bar jumps and the wave wiggles, proving the mic works. The active audio track’s getSettings() and the AudioContext report the human-readable device name (track.label), sample rate, channel count and reported latency, and enumerateDevices() lists every audio input (labels appear once permission is granted) to populate a microphone selector; choosing another mic stops the current tracks and restarts getUserMedia with { deviceId: { exact } } so headsets and USB mics are easy to test. The Record button wraps the MediaRecorder API around the live stream — ondataavailable collects chunks, onstop assembles a Blob — which becomes an object URL played back in an audio element so you can hear yourself; clips can be downloaded (.webm/.ogg), discarded or kept several at a time, and every object URL is revoked on discard, clear-all and unmount. The level meter surfaces clear NO SIGNAL / GOOD / TOO LOUD states with a “speak and watch the bar move” hint, and the Start/Stop and Record controls repeat in a sticky mobile bar. Every error path is mapped to a friendly message — permission denied (NotAllowedError/SecurityError), no mic (NotFoundError/OverconstrainedError), mic in use by another app (NotReadableError/TrackStartError), insecure context, and unsupported browser — and a track ‘ended’ listener catches a mic unplugged or taken over mid-session. Your microphone audio is analysed only on the device and is never uploaded — there is no server or account — and any recording is an in-memory blob that stays in your browser, written to disk only if you download it; the mic track is stopped, the AudioContext closed, the recorder stopped and object URLs revoked the instant you press Stop test, leave the page or unmount, turning the mic indicator off. Honestly, the sample rate, channel count and latency shown are what the browser and device negotiated for this session rather than the hardware’s maximum, the dBFS figure is a relative loudness reading rather than a calibrated SPL measurement, a flat level meter usually means the wrong mic is selected, the mic is muted, permission is missing or another app (Zoom, Teams, Meet, OBS or another tab) holds it, and access requires permission and a secure (https) context, with recording relying on the MediaRecorder API that a few older browsers lack.