Before you start
Choose one input. Disconnect or mute unused devices if their similar names make selection unclear, then note the microphone you intend to use. Place it at the distance and angle you will use during the call.
Choose a short sentence and keep it unchanged for each recording. The goal is a comparable sample, not a polished performance. Changing words, distance and input volume together makes the result harder to interpret.
What needs to pass? The stream must start, normal speech must move the meter without repeated clipping, and local playback must remain understandable without unexplained silence. Those are browser observations, not a complete rating of microphone quality.
Run the check
Press Start and allow access. Choose the intended device from the input selector after labels appear. Speak the fixed sentence at the planned distance, then stay quiet long enough to see how the waveform and displayed floor change.
Record 5s locally and repeat the same sentence. Wait for recording to finish, then play the sample once with live monitoring off. Headphones prevent the playback from returning to the microphone and confusing the next capture.
Does the meter move but playback sound faint? Record the average and peak before changing anything. Adjust one supported input-volume or distance setting, start a fresh run and repeat. Keep output volume unchanged while comparing captured loudness.
Read your result
A moving waveform and finite average show that samples reached this page. Peak near the digital ceiling together with repeated clipping suggests that the delivered signal is too strong for this setup. The display cannot show analogue headroom before conversion.
Quiet gaps mean exact digital silence returned after activity for at least the tool's 120 ms review window. Listen for the same interruption in playback. A gate can produce silence by design, so the count alone does not identify a failing cable or driver.
Can you understand every word at normal playback volume? Use that question with the meter. A louder waveform is not automatically clearer, and a clean waveform cannot reveal how a call service will compress or suppress the input later.
Measurement limits
The page analyses a stream after device, operating-system and browser processing. It cannot measure microphone sensitivity in volts, acoustic sound pressure, frequency response or the complete delay to another participant. Keep claims limited to this captured browser sample.
Auto gain, echo cancellation and noise suppression may differ between this tab and a call application. The browser track setting can expose some negotiated values, but it does not provide a full account of every driver or application process.
The local recording remains in memory and is not uploaded. Device labels appear after permission to support input selection. If privacy or input access is blocked at the operating system, this page cannot override that setting.
What to do next
Keep the baseline if speech is clear, levels respond and playback has no repeatable gaps or clipping. Then verify the same named input in the call application. Its own preview is the final check for that application's route and processing.
If input is quiet, compare placement before adding gain. If it clips, reduce the input level or increase distance, then repeat the fixed sentence. Make one change per run and keep the result that improves playback without creating another problem.
If the stream does not start or the intended device is missing, stop the content test and diagnose access. Follow the microphone not working guide through system input, permission and application selection before considering model-specific support.
Symptom table
| Symptom | Likely layer | Next action |
|---|---|---|
| Level never moves | Wrong input, blocked permission or inactive track | Select the input and request access again |
| Strong input clips | Gain or distance is too high | Lower input gain or increase distance |
| Playback has gaps | Digital silence or processing interruption | Compare a local recording with processing changed |
Common questions
What level should my microphone show when I speak normally?
There is no universal dBFS target for every voice, distance and application. Look for a responsive average, peaks below repeated clipping and understandable local playback. Record the baseline, then compare one setting change. The page does not convert its digital level into acoustic loudness at the microphone.
Why does my microphone sound fine here but bad in calls?
The call app may select another input or apply its own gain, compression, echo cancellation and noise suppression. Confirm the same device name and run the app's check. This local sample establishes what reached the browser; it cannot predict the audio encoded and sent by another service.
Can a browser mic test damage or change my microphone?
This test requests a stream and reads samples. It does not write firmware or change physical gain controls. You may choose to change a system input setting during comparison, so record its original value first. Stop the stream or close the page when the check is complete.