Audio latency test

Round-trip latency is the full loop a sound takes through your computer: out through the output buffer, the digital-to-analog converter, and the speaker, across the room as actual sound, then back in through the mic, the analog-to-digital converter, and the input buffer. It is the delay you notice when you monitor your own voice or when audio drifts behind video.

The test measures that loop directly. It schedules a short 1 kHz click at a known moment on the browser's audio clock while recording your mic. First it reads your room's noise floor in the 100 milliseconds before the click, then it looks for the first sustained burst of energy after the click that clears that floor by a wide margin. The time between the scheduled click and the detected echo is your round trip. Browser echo cancellation and noise suppression are switched off during the test, either one would erase the very echo being timed.

This is also why speakers are required. On headphones the click goes straight into your ears and never reaches the mic, so there is no echo and nothing to time. The test detects that case and says so instead of showing a made-up number. If you mostly care about the delay you hear while monitoring yourself, start with what monitoring latency actually is.

As a rule of thumb for a monitoring chain: under 20 ms feels tight, your voice reads as immediate. 20 to 50 ms is usable for streaming, calls, and screen recording, though you would notice it monitoring a vocal. Above 50 ms you feel it: your own voice comes back as a distracting slap, and lip sync starts to drift.

One honest caveat before you judge your hardware: this tool measures your browser's full loop, which stacks your hardware, the operating system's shared mixer, and the browser's own buffers on both sides. So treat the reading as an upper bound, not as what a dedicated low-latency driver gets on the same machine. A laptop that reads 80 ms here can still monitor far faster through ASIO. The browser-reported output buffer shown under your result gives a hint of how much of the total is the browser's own output side. If your everyday monitoring feels slow, the practical fixes live in how to fix mic monitoring latency.

For scale: patchd's engine measures about 10.7 ms through the bus at a 512 buffer at 48 kHz, and smaller ASIO buffers take it lower. That number comes from the buffer math itself, latency in milliseconds is buffer samples divided by sample rate. patchd has no buffer picker, your audio interface's own control panel sets the buffer and patchd reads the result live in its master clock readout.