DevicedialAbout
Devicedial answers one kind of question: is this thing actually working properly? The microphone, the camera, the keyboard, the mouse, the controller, the screen. Each gets its own test, and each test ends in a number with a unit โ not a tick.
Why it exists
There is no shortage of device testers. There is a shortage of ones that tell you anything. Type “mic test” into a search engine and most of what comes back draws a moving bar and says “your microphone is working” โ which is true of a microphone recording you 14 dB too quietly, and true of a mouse reporting at a fifth of its advertised rate.
That is the gap: a tick answers “is it connected”, and nobody has ever searched for that. People arrive here because something is wrong and they cannot tell whether it is the hardware, the driver or the app. That question needs a measurement, a scale it sits on, and a sentence saying whether the figure is normal. All three, every time.
Where the numbers come from
Nothing is estimated and nothing is looked up from a database of devices. Microphone level is read
from a Web Audio analyser on the live stream and reported in dBFS โ decibels relative to full scale,
where 0 is the loudest a digital sample can be and every real signal is negative. Rates and times come
from requestAnimationFrame timestamps and pointer-event timing, so a refresh rate is the rate of
frames the browser actually delivered. Rollover is counted from key events, controller axes from the
Gamepad API’s โ1 to 1 range, resolution from the screen the page is on.
Where a figure needs a definition, the page gives it: a word is five characters including the space, so typing speed does not depend on which words you type; the drift score is our own scale from the resting deflection of a stick, because the Gamepad API reports a ratio and not an angle. Each tool explains its own arithmetic under How we calculate, with a worked example computed by the same code that runs in your browser.
What this is not
A browser measures what the browser was handed, and that is an honest limit rather than a disclaimer. Mouse DPI cannot be read by a web page at all โ a site that prints yours is guessing. Neither can a panel’s native sampling rate, the firmware behind a driver, or whether an app downsampled a camera stream before the page saw it. Those tests say so instead of inventing a figure.
The tests also do not diagnose hardware faults on their own. A reading that looks wrong narrows the question down to a device, a cable or a setting โ which is usually all you needed, and occasionally the point at which you take it to somebody with a screwdriver.
Nothing is uploaded. Audio, video, keystrokes and clicks are measured frame by frame and discarded; no test makes a network request while it runs, and no account is needed for any of them.
Found a reading that looks wrong, or a test we are missing? Tell us โ corrections get made and requests get read.