Mic Test Online — Free Mic Check

This is the fastest complete microphone check you can run: the page requests your mic through getUserMedia() (the W3C Media Capture and Streams API), draws its live waveform, meters the signal in dBFS, and gives you a headphone monitor toggle so you can hear exactly what your input is transmitting before a meeting, lesson, stream, or recording session.

How to use it

1. Press Start Test and grant the permission prompt. The browser asks first; on macOS, the system shows a second dialog the first time a browser app requests microphone access. Both are required. If you clicked “Block” once, the prompt will not return until you clear the permission from the site settings panel (the lock or tune icon at the left of the address bar).
2. Select the intended device from the input list. Device names repeat constantly (“Default”, “Microphone (USB Audio Device)”, “Stereo Mix”). The one that matters is whichever hardware your cable is actually plugged into; laptop mics, webcam mics, and interface inputs can all appear at once.
3. Put headphones on, then enable “Hear yourself” monitoring. The monitor routes the capture stream to your output with only Web Audio buffer delay between them. Never enable it with speakers live: the mic-to-speaker loop hits a howling feedback squeal in under a second.
4. Speak normally, then test the hard cases. Count from one to ten at conversational volume, say a few plosives (“popcorn”, “baseball”), and laugh. Laughter is where mis-gained mics clip.
5. Watch the waveform and the level meter together. Voice peaks should land between roughly −18 and −6 dBFS, and the trace should return to a thin noise line between words. Flat-topped wave crests mean the converter is saturated.
6. If it fails here but works elsewhere (or vice versa), the problem is routing, not hardware. This page reads the browser-visible device; the conferencing app may be choosing a different one. Check the app’s own audio settings after confirming a clean signal here.

What the meters are confirming

A useful mic check verifies three distinct properties, and this page is built around exactly those three.

Signal presence and routing. The first job is proving the OS is feeding the browser the mic you think you’re using. A muffled, distant-sounding signal while an external mic is “selected” is the classic symptom of the system’s default communication device still pointing at the built-in array.

Waveform shape. Clean speech is an irregular stack of sharp attack transients followed by decaying oscillation, with visible symmetry above and below the center line. A squared-off or flattened trace is digital clipping, which is permanent damage to the recording, not a display artifact.

Gain staging headroom. Voice has a crest factor of 12–18 dB between average and peak, so peaks that already touch −3 dBFS in normal speech will slam the 0 dBFS rail the moment you raise your voice. Target peaks of −18 to −6 dBFS: audible, well above the noise floor, still 6+ dB from the cliff.

Level readout quick reference

What the meters do	What is happening	Fix
No trace at all, meter at −∞	Wrong device, hardware mute switch, or blocked permission	Re-select input; check physical mute; reset site permission
Trace moves only below −40 dBFS	Mic too distant or OS input slider down	Move to 15–20 cm; raise input level in OS sound settings
Peaks −18 to −6 dBFS, thin baseline between words	Correctly gain-staged speech	You are ready; stop here
Meter flashes above −3 dBFS or tops flatten	Input gain too high; clipping on every plosive	Lower OS/interface gain two steps, then re-test the laugh
Loudness pumps up and down on its own	Automatic gain control is engaged	Disable “audio enhancements” (Windows) or Voice Isolation (macOS)

Permissions, privacy and device quirks

getUserMedia() only runs in a secure context (HTTPS), so an insecure or proxied network can block the request outright, usually with a generic permission error. Both Chrome and Firefox refuse mic access on plain-HTTP pages by design.

Windows separates microphone permissions into “Store apps” and “Desktop apps” lists under Settings > Privacy & security > Microphone; a browser that is on the system allowlist still needs the desktop-app switch. macOS shows an orange dot in the menu bar whenever any process is actively capturing; if the dot glows while this page is closed, something else holds your mic.

Bluetooth headsets carry two profiles. As an output they stream high-quality stereo (A2DP); the instant an application opens their microphone, they drop to the hands-free profile (HFP) and audio quality falls off a cliff, at roughly 8–16 kHz speech bandwidth. If your music suddenly sounds like a phone call, an app is holding the headset mic open.

On Windows, a running conferencing app can also take the device exclusively; quit Zoom, Teams, or OBS if this page sees a dead input that other apps report as working.

All capture here is transient: audio buffers are analyzed in the page’s local memory and discarded each frame, never encoded or sent anywhere.

Related tools

• Microphone Quality Test — The deep measurement: noise floor, SNR, and a 0–100 quality score.
• Mic Loopback — Record ten seconds of yourself and play it back exactly as listeners will hear it.
• Audio Latency Test — Measure the output delay of the same headphones in milliseconds.
• Speaker Test — Prove the output side of the machine before blaming the input side.