DevicedialMic, sound and camera tests
The hardware people notice only when a call goes wrong. Each test here gives you the reading, not a green tick โ and tells you whether the problem is the device or the app on top of it.
Camera test
Two numbers the camera reports about itself, and what each one is costing you.
Echo test
Five seconds in, five seconds back โ and the room comes with it.
Headphone test
Two of the three faults are obvious. The third only shows when you switch between them.
Mic test
Speech should peak inside a band most people have never been told about.
Speaker test
Left, right, or both โ and seven frequencies that go missing for different reasons.
Webcam mirror
A screen makes a better light than a bathroom mirror ever gets. The lens is the part that lies.
These are the devices that only get tested in public. Nobody checks a microphone before the meeting; they find out during it, from six people saying they cannot hear anything. The point of this group is to move that discovery ten minutes earlier and to make it specific โ not “my mic is broken” but “my peak is โ26 dBFS, so I need the input gain up”.
The order that answers most problems is level, then channels, then loop. The mic test tells you whether your voice is arriving loud enough. The speaker test plays each side on its own so a dead channel cannot hide behind the working one, and the headphone test adds the leftโright and phase checks that matter once both drivers sit against your head. The echo test closes the loop by showing what your microphone picks up from your own speakers, which is the real cause of nearly every complaint about echo.
The camera pages work the same way. The camera test shows the live picture with the resolution and frame rate the browser actually negotiated, which is regularly lower than the box claims and drops further in dim light. The webcam mirror is the plain full-window view for the practical jobs โ checking framing, lighting and what is visible behind you before you turn the camera on somewhere that matters.
Frequently asked questions
Which of these should I run before a call?
Two, in this order. The mic test tells you whether your voice is arriving at a usable level โ peaks around โ12 to โ6 dBFS โ and the speaker test tells you the other person will be audible on both channels. If you wear a headset, add the headphone test, because a channel that has quietly died is almost impossible to notice until someone speaks only into that ear.
The device works here but not in my meeting app. What now?
The problem is device selection or permission, not hardware. Every browser and every app keeps its own idea of which microphone, speaker and camera to use, and a headset typically shows up as two or three separate devices. Pick the input and output by name in the app rather than leaving them on Default, then confirm the app has microphone and camera permission in your operating system’s privacy settings.
Why does the camera test show a lower resolution than my webcam claims?
Because the browser negotiates a stream rather than taking the sensor’s maximum. It settles on a size and frame rate that suit the current lighting, the USB bandwidth available and whatever the page requested, and low light in particular makes cameras drop frame rate to keep the image usable. The number to watch is not the peak the box quotes but the resolution and frame rate you are actually sending right now.
Does any of this record me?
No. Audio and video are read as a live stream, measured frame by frame and discarded โ nothing is written to disk, nothing is sent to a server, and there is no account. The browser’s own indicator shows while a stream is open, and revoking permission in the padlock menu next to the address ends it immediately.
Why do I hear an echo of myself on calls?
Almost always because your speakers are feeding your microphone. The remote sound comes out of the speakers, the mic picks it up and sends it back, and the other person hears themselves a moment late. Headphones end it outright; if you cannot use them, lower the speaker volume, move the mic away from the speakers, and check that the app’s echo cancellation has not been switched off โ several apps disable it automatically when they detect a high-quality input.
How the test works
AnalyserNode attached to the live input stream and is
reported in dBFS, decibels relative to full scale: 0 is the loudest value a digital sample can
hold, so every real signal is negative and quieter means further below zero. Speaker and headphone
tests generate tones in the browser itself and route them to one channel at a time, which is what
makes a dead side or a swapped pair obvious; the phase check plays the same tone in both channels
with one inverted, and a signal that collapses in the middle instead of sounding wide means the
wiring is out of phase. The camera test reads the resolution and frame rate the browser negotiated
for the live track, not the maximum the sensor advertises. Nothing is recorded, uploaded or
stored โ every measurement happens in the tab and disappears with it.Sources: