Why does the mic test disable noise suppression and auto gain?
Disabling browser digital signal processing lets you hear how your microphone hardware and room actually sound before video-call software masks defects. Calling getUserMedia with echoCancellation, noiseSuppression, and autoGainControl set to false prevents the browser from artificially boosting whispers or gating fan hum.
Most online mic check tools leave default browser filters turned on, which hides low input gain, electrical buzz, or aggressive room echo until an important meeting starts. Turning those filters off gives an honest baseline for your headset, USB condenser, or laptop array. Before joining a video call, you can also check your camera, resolution and FPS to verify framing alongside audio.
The requested audio constraints follow the W3C Media Capture and Streams Specification, which exposes hardware capabilities through standard track settings. Reading those properties immediately after permission is granted reveals whether the operating system opened your intended studio mic or fell back to a Bluetooth hands-free profile.
Disabling browser audio processing for true diagnostics
When autoGainControl is active, a quiet microphone looks normal on a meter because the browser cranks digital gain up between words and amplifies background hiss. Leaving it off ensures the meter reflects true analog and operating-system input volume.
Similarly, turning off noiseSuppression exposes air-conditioner rumble, keyboard clatter, and desk vibration that might otherwise clip consonants. Once you know the raw signal is clean, conferencing apps can apply their own noise filters on top of a healthy input.
Reading sample rate and channel metadata
The Details sidebar inspects MediaTrackSettings on the active audio track as soon as the stream opens. It displays the hardware Device label, Sample rate rounded to one decimal place in kHz, and Channels formatted as Mono, Stereo, or a numeric count.
If a multi-microphone setup is connected, the Microphone dropdown filters out synthetic default entries so every option maps to a distinct physical or virtual input. Switching devices restarts the live analyzer immediately without reloading the page.
How to test your microphone online with the level meter and clipping check
Real-time audio diagnostics run through a Web Audio AudioContext connected to an AnalyserNode configured with a 2048-sample FFT buffer. Because the analyzer reads samples passively without connecting to audioCtx.destination, you can test my mic levels freely with speakers on and zero live echo.
Every animation frame pulls time-domain floating-point samples as documented in the MDN Web Audio AnalyserNode API reference. The oscilloscope canvas scales to device pixel ratio and draws every fourth sample across the waveform width so speech transients stay crisp at 48 kHz or 44.1 kHz sample rates. When you are ready for full-length takes, you can record audio from your microphone directly in your browser.
| Verdict state | Measured threshold | Meter appearance | Recommended action |
|---|---|---|---|
| Too loud: clipping | Peak amplitude > 0.99 |
Danger red above 0.92 |
Step back or lower OS input gain |
| Quiet / muted prompt | Loudest RMS < 0.25 |
Low muted fill bar | Check physical mute switch or input device |
| Quiet warning | Loudest RMS 0.25 to 0.45 |
Mid-low muted bar | Move closer or raise system mic volume |
| Good level | Loudest RMS >= 0.45 |
Solid dark ink bar | Ready for calls and recordings |
Decibel scaling from minus 60 dB to 0 dB
Raw root-mean-square (RMS) voltage is logarithmic to human hearing, so linear meters make normal speech look barely active. The level helper converts RMS into decibels using 20 * log10(rms) and maps the -60 dB to 0 dB range linearly onto a 0 to 100 percent meter scale.
Under that curve, conversational speech sits comfortably between 45% and 92% of the role="meter" bar before turning red. Background silence stays near the bottom edge unless excessive fan noise or electrical hum pushes the noise floor upward.
Peak hold tracking and clipping detection
A thin vertical marker tracks recent loudest levels using a per-frame decay factor of 0.97, letting you glance away while speaking and still see where peaks landed. Any single sample exceeding 0.99 amplitude sets the sticky clipping flag so brief plosives on letters like P and B do not go unnoticed.
If the verdict reports clipping, reduce input gain in your operating system sound settings or angle the microphone slightly off-axis from your mouth. Once clipping clears, click Stop and start again to confirm the loudest run stays in the healthy range.
Recording a 5-second playback clip without speaker feedback
Visual meters confirm signal presence, yet only listening to your own voice reveals room reverb, sibilance, or muffled headset capsules. Clicking Record 5 seconds and play back captures a short local clip using the browser MediaRecorder interface so you can judge real-world clarity.
The recording pipeline follows the MDN MediaRecorder API and stops automatically after 5 seconds (5,000 ms) via a built-in timer. Once the clip finishes, an inline HTML <audio> player appears below the controls. If you want to compare your voice against synthetic narration or check speaker output separately, try a read aloud tool on the same device.
Tip: Say a full sentence with normal breathing pauses during the 5-second test rather than tapping the desk. Hearing how your room decays between words tells you far more about call quality than a single clap.
Automatic live mic shutoff during playback
Playing recorded audio through laptop speakers while the microphone remains active creates an immediate feedback squeal. To prevent that, the <audio> element attaches an onplay handler that calls stop() on the live microphone stream the instant playback begins.
Any previous playback object URL is revoked from memory whenever a new clip finishes or when the component unmounts. Nothing is written to disk or persisted across page reloads unless you save it yourself.
Audio formats supported across browsers
Before starting MediaRecorder, the helper probes MediaRecorder.isTypeSupported in priority order: audio/mp4 with AAC (mp4a.40.2, .m4a), audio/webm with Opus (.webm), and audio/ogg with Opus (.ogg). If a browser rejects explicit MIME options, it falls back to the browser default MediaRecorder container cleanly.
Why is my microphone not working in the browser?
When no audio reaches the meter, the cause is almost always a blocked site permission, an operating-system privacy toggle, or another app holding exclusive control of the microphone. Modern browsers also block getUserMedia completely on insecure HTTP pages, requiring HTTPS or localhost to expose audio hardware.
The MDN MediaDevices.getUserMedia() API raises specific DOMException names that the tool translates into plain-English fixes. A NotAllowedError or SecurityError means browser or OS permission is blocked, while NotReadableError indicates another video call or tab has locked the device. After confirming your voice sounds clear, you can also listen to documents hands-free with PDF to audio using the same speakers or headphones.
If NotFoundError appears, verify the USB cable, Bluetooth pairing, or system sound input list before clicking Test my microphone again. Testing here first takes seconds and catches muted headsets, wrong default inputs, and clipping gain before your next meeting or stream begins.