Tools /  Audio Latency Test

Audio Latency Test

音声遅延測定
検証済TESTED
⚠ Protect your ears: start at low volume. The metronome ticks play at a modest level; check your system volume first.
BASE LATENCY
·
OUTPUT LATENCY
·
TAPS
0
PERCEIVED OFFSET
·
METRONOME 90 BPM

Start the metronome, then tap the pad in time with what you hear.

Negative offset = anticipating the beat. This measures your whole chain (output latency plus your own timing bias), so compare readings (Bluetooth vs wired) rather than trusting the absolute number; output latency may read n/a where the browser doesn't expose it.

How it works

仕組み

Measure how much delay sits between your browser and your ears. The tool reads the audio stack’s own numbers first (AudioContext baseLatency and outputLatency), then runs a hands-on check: a metronome ticks at 90 BPM and you tap along, and after eight taps it reports your mean offset from the beat in milliseconds. Latency depends entirely on your device, driver, and output. Wired headphones commonly sit under 30 ms while Bluetooth can add 100-300 ms, so no spec sheet can tell you what your setup actually does. Rhythm-game players and musicians, this one’s for you.

What do baseLatency and outputLatency actually mean?

baseLatency is the delay the browser’s own audio pipeline adds (typically 3-12 ms), while outputLatency estimates the rest of the trip through the OS mixer, driver, and hardware to actual sound. Not every browser reports outputLatency (it may read n/a or 0), and it often only settles once audio is playing. Their sum approximates the true click-to-sound delay, but the tap test captures things the numbers miss.

Why is my perceived offset negative?

A negative offset means your taps land before the scheduled tick: you’re anticipating the beat, which is completely normal for a steady rhythm since the brain predicts rather than reacts. Trained musicians often sit slightly ahead or dead-on, while high output latency pushes the number positive because you hear each tick late. That’s why the relative change matters more than the raw value.

How can I use this to compare Bluetooth and wired audio?

Run the tap test once on each output, keeping everything else identical, and compare the two offsets; the difference between them is a decent estimate of the extra delay your Bluetooth link adds. Expect roughly 100-300 ms more depending on codec (AAC and SBC are slower; aptX Low Latency and LE Audio are much better). For rhythm games and recording, wired remains the safe choice.

More on the bench 他の道具
Hearing Frequency Test
Stereo Channel Test
Mic Level Meter
Anti-Ghosting & NKRO Tester
Reaction Time Test
Multitouch Tester
Color Contrast Checker
Palette Accessibility Auditor