Why this site exists
Audio Delay Test helps you explore how consistently you tap along with a repeating sound. You choose a tempo, listen to a few warm-up beats, and tap a pad or press Space. The result describes your timing in that session. The tool is designed for people who want to observe their rhythm, repeat a simple comparison, or understand why a sound feels difficult to follow.
The word latency covers several different problems. A headphone's output delay, a recording system's round trip, and a person's anticipation of a beat are related but distinct. We keep that distinction visible so a convenient browser number does not become a misleading hardware claim.
What you can observe
The beat timing test shows a signed tap phase offset after at least ten valid taps. Positive values mean your taps fall later than the nearest browser output reference. Negative values mean they fall earlier. A separate spread value, the median absolute deviation, describes how tightly the taps cluster.
The display uses up to 32 recent valid taps. It labels segments stable or unstable under the tool's consistency rules and shows a chart of the signed observations. Stability helps you assess repeatability; it does not remove human anticipation or certify the accuracy of a playback device.
How the browser test works
An original click is scheduled on one Web Audio context. Browser timers replenish a queue of future audio starts; they do not directly set the beat clock. Your input timestamps are compared with scheduled beats through validated browser output timestamps. When the required mapping is unavailable or unstable, the tool leaves the phase result unavailable.
At 120 beats per minute, the interval is 500 milliseconds. Phase is circular: a tap 300 milliseconds after one click is also 200 milliseconds before the next. The signed result therefore describes a relationship within one beat. It cannot identify extra delay spanning a whole beat. The result guide gives worked examples.
Make useful comparisons
Keep the browser, tempo, input method, and listening conditions consistent when comparing sessions. Change one playback path at a time and complete several short runs. Note both the median and spread, rather than treating one unusually large or small value as a device specification.
Pause starts a fresh segment when you resume. Switching away from the page, suspended audio, and detected clock or scheduling interruptions also invalidate a segment. These boundaries keep separated observations from being silently combined. After changing your system output, start a new run even if the browser did not detect the switch.
Keep the limits in view
This site does not identify a Bluetooth codec's exact delay, measure sound arriving at your ear, record an acoustic round trip, or measure picture-and-sound alignment. Those tasks need a different method and, for physical timing, an appropriate reference setup. A very steady tapping result still includes your own synchronization and input path.
There is no universal good or bad headphone score here. The tool does not diagnose hearing, motor function, driver faults, or professional equipment suitability. Read the Disclaimer before drawing conclusions from a result.
A test without audio recording
The beat test does not ask for microphone access, accept audio uploads, or require an account. Its sample array stays in the current tab's memory and clears on reload. Stop can retain the completed segment on screen until you start again or reset; there is no saved device history.
We use Google Analytics and Microsoft Clarity to understand visits and page interactions. Website usage information is separate from the local timing calculation. The Privacy page explains test data, browser storage, and these third-party services.
Explain the method, not a bigger promise
The site's explanations are based on its implemented controls, reproducible arithmetic, and browser API documentation. AI assisted the implementation and explanatory writing. No laboratory testing credentials, expert endorsement, or measured headset dataset are claimed. Examples illustrate the calculation; they are not reports of hardware trials.
If you find an unclear instruction, inaccessible control, or result explanation that does not match what the page does, use Contact to describe it. Include the page, browser, playback path, and steps involved so the issue can be understood without guessing about your setup.