Microphone Test

No. The recording exists as raw numbers in your browser’s memory, and nothing in this page can send it anywhere — the only network request the tool makes is a single message saying the word “mic-test”, so that the usage counter is honest. Nothing is written to your disk either, so there is no file to find or delete afterwards. The simplest way to satisfy yourself: turn off your wifi and run the whole test, playback included. It works identically.

Because the most common microphone complaint is not “it does not work” — it is “people say I sound bad”, and no meter can answer that. Hearing five seconds of yourself played back is what turns “my mic is fine” into “oh, I sound like I am in a bathroom”. A page that only shows you a moving line can tell you a signal exists. It cannot tell you what that signal sounds like to everyone else.

Because of how Bluetooth works, and it is not a fault in your earbuds. While you are only listening, the connection carries high-quality stereo. The moment something needs the microphone, the link switches to a hands-free mode that is mono and drastically narrower — old narrowband profiles cut everything above roughly 4 kHz, the newer wideband ones above roughly 8 kHz. Speech runs well past both. This test measures where your audio actually stops and tells you which of the two you are on, which is why the number it reports can be so much lower than you expect. Every wireless pair does this; how bad it sounds varies enormously between them.

Two places. In this page, the picker above the Start button appears once you have run the test once, and switching it re-runs the test on that device. System-wide, it is Settings then System then Sound on Windows, or System Settings then Sound then Input on a Mac. Note that browsers keep their own default separate from the system one, which is exactly why someone can plug in a headset, see it connect, and still be heard through the laptop’s built-in microphone.

Almost always because the other application is pointed at a different device, or because it is still holding a microphone from a call you thought had ended. If this test passes, the hardware, the driver and the operating system are all fine, and the problem is inside that application’s own audio settings. If this test cannot open the microphone at all and says another app is using it, close every call app and voice recorder and try again — several of them keep the device even while muted.

Speech averaging around −20 dB with peaks below about −3 dB is comfortable. Below −36 dB you are quiet enough that noise gates in call software will start clipping the beginnings of your words off. Above −3 dB you are close enough to the ceiling that loud syllables square off and distort. What matters more than the number is the gap between your voice and the room behind it: about 15 dB is the point at which the other person stops noticing your fan.

Because half of what makes someone hard to listen to is not their voice, it is everything under it. Measuring the room on its own for a couple of seconds gives a noise floor, and the difference between that and your speech is the number that predicts whether people will ask you to repeat yourself. It is also the only way to catch the case where a microphone is technically working but sits next to a laptop fan.

No, and it does not pretend to. Judging microphone quality means comparing it against a known reference in a treated room; a browser has neither. What this can do is measure what actually arrived — your level, the room, distortion, the frequency range — and name a specific fault when one is present. That is a different and more useful question than a score out of ten, because it comes with something to do about it.