Audio latency test
Read the base and output latency your browser's audio engine reports, understand what the total means in practice, and see which changes actually reduce it.
Last reviewed on August 9, 2026
Read the base and output latency your browser's audio engine reports, understand what the total means in practice, and see which changes actually reduce it.
Last reviewed on August 9, 2026
Run the measurement to see your browser's reported audio delay.
The test reads the two latency figures your browser's audio engine reports: base latency, the buffer the browser holds internally, and output latency, the browser's estimate of the time between handing a sample to the operating system and it leaving the speaker. Added together, they are the floor of your system's audio delay.
They are a floor, not a total. Not included: the Bluetooth link (typically 100–200 ms), any delay in the microphone path, processing inside a meeting app, display buffering, and network transit on a call. A machine reporting 12 ms here can still have 250 ms of real end-to-end delay on a wireless headset.
No permission needed. This test only reads numbers the browser already exposes. It does not open your microphone.
| Reported latency | How it feels | Good enough for |
|---|---|---|
| Under 10 ms | Imperceptible | Everything, including live monitoring while recording |
| 10–20 ms | Effectively instant | Music production, competitive gaming |
| 20–40 ms | Fine unless you are listening for it | Calls, gaming, general use |
| 40–80 ms | Noticeable when watching lips move | Calls and media; awkward for live monitoring |
| Over 80 ms | Clearly delayed | Passive listening only |
Useful reference points: about 20–30 ms is where most people begin to notice audio trailing video; roughly 10 ms is the threshold at which musicians find live monitoring unplayable; and standard Bluetooth adds far more than either figure, which is why wireless earbuds are unsuitable for recording.
Run the test in two browsers on the same machine and you will usually get two different answers. That is expected:
Because of this, compare the same browser across configurations rather than comparing browsers against each other.
In roughly descending order of effect:
The full explanation of where each millisecond comes from is in audio latency explained.
Three common complaints trace back to delay, but each has a different cause:
If you are chasing an echo specifically, run the recorder: an echo that shows up in your own local recording is a room-and-microphone problem, while one that only others hear is a routing problem.
Under 10 ms is imperceptible, 10 to 20 ms is effectively instant and fine for music production, 20 to 40 ms is fine for calls and gaming, and over 80 ms is clearly delayed.
No. It reports the browser's internal buffering and its estimate of output delay. It excludes Bluetooth, drivers, meeting-app processing, display buffering, and network transit.
Chrome and Edge report both base and output latency, Firefox reports them inconsistently across platforms, and Safari often exposes neither, in which case the test falls back to a theoretical minimum. Compare the same browser across configurations rather than comparing browsers.
Typically 100 to 200 ms with standard codecs — far more than everything else combined, and the reason wireless headphones are unusable for recording while monitoring yourself.
Use a cable instead of Bluetooth, plug in directly rather than through a hub or dock, close other audio applications, disable audio enhancements and spatial audio, and match sample rates between the OS and your applications.
No. It only reads values the browser already exposes, so no prompt appears.