Audio Latency Test — Bluetooth ms

Audio latency is the delay between the instant your browser schedules a sound and the instant that sound physically arrives at your eardrums. This tap test measures the complete output chain (OS mixer, wireless codec, headphone DSP) in milliseconds, so you can tell whether the lag you notice lives in your Bluetooth link, your device, or your app.

How to use it

1. Route output to the device under test. Pair the Bluetooth headphones or plug in the interface, then make it the active system output. Testing a device the OS is not sending audio to measures nothing.
2. Start at moderate volume. Set system volume near 30%. The clicks are short 1 kHz bursts, but an unfamiliar headphone fed a full-scale transient at reference volume is unpleasant.
3. Press Start. The page schedules one click every 1000 ms on the Web Audio hardware clock, which runs at the sample rate and does not drift with browser timer throttling.
4. Tap in time with each click you hear. Keep watching the button and tap at the rhythm you perceive, not the rhythm you imagine. Ten taps complete a trial.
5. Read the median offset. Each tap is compared with its scheduled click, wrapped into a ±500 ms window so early and late taps never cancel each other out, and outliers are discarded. The reported median is your output latency in milliseconds.
6. Change one variable and re-run. Wired versus wireless, one codec versus another, one pair of earbuds versus another. A 150 ms gap between two runs is a real difference. A 10 ms gap is human noise.

Where audio latency comes from

A click travels through up to five delay stages before it reaches you:

1. Operating system audio buffers (5–20 ms): the OS queues samples in ring buffers, typically 128 to 1024 samples. At 48 kHz that is 2.7 to 21.3 ms of stored time per buffer stage, and playback plus capture usually doubles the cost.
2. Codec encoding and framing (15–40 ms): lossy wireless codecs group samples into analysis windows before compression. An AAC frame holds 1024 samples, about 21.3 ms at 48 kHz, and the decoder needs a second frame in hand before it can start.
3. RF transmission and retransmissions (20–100 ms): Bluetooth shares the 2.4 GHz band with Wi‑Fi. Every packet that fails must be resent, so receivers hold a safety buffer deep enough to survive interference bursts.
4. Receiver buffering and digital-to-analog conversion (5–20 ms): the headphone’s own queue and DAC add a fixed block delay before anything moves the driver.
5. DSP and active noise cancellation (5–15 ms): ANC, transparency, and spatial-audio filters all need future samples to work on, which is added delay.

The browser itself is usually the smallest term. Web Audio renders in 128-sample quanta (2.7 ms at 48 kHz), and AudioContext.baseLatency and outputLatency report how much the platform stack contributes, so you can cross-check the tool’s number against what Chrome or Safari claims.

Typical latency by connection type

Measured results vary by phone, firmware, and RF environment, but the ranges below are what you should expect a functioning system to land in.

Output path	Typical end-to-end delay	Dominant source of delay
Wired analog (3.5 mm) with OS default buffers	5–30 ms	OS ring buffer size
USB class-compliant interface, 128–256 sample buffers	5–15 ms	Two render quanta plus converter
Classic Bluetooth aptX Low Latency (both ends support LL)	32–40 ms	Fixed short-frame codec pipeline
Bluetooth LE Audio (LC3)	20–60 ms	Configurable frame size and retransmission budget
Bluetooth AAC, iOS with AirPods	130–170 ms	AAC framing plus conservative receiver buffering
Bluetooth SBC, Android default	150–220 ms	Wide interference margin, large decode buffer

Use these numbers against the practical budgets the industry already publishes. ITU-T G.114 calls 0–150 ms one-way acceptable for conversational voice, which is why calls over headsets feel fine even though the number looks bad. For music monitoring and gaming you want the total under about 20–40 ms. For video, ITU-R BT.1359-1 found viewers reliably detect sound leading picture by roughly 45 ms or trailing it by roughly 125 ms, and streaming players slip the video track behind by that much automatically, which is why your 180 ms earbuds play YouTube in sync while a game cannot.

Accuracy, limits and troubleshooting

The measurement fuses machine and human. Tapping a beat involves motor reaction and anticipation; with a regular 1000 ms pulse, most people lock in within about ±20–30 ms. The median filter and outlier removal converge on the true latency within roughly ±10–15 ms over ten taps. For millisecond-exact figures you need a loopback measurement or the DAW’s driver panel, not a thumb.

Background tabs distort timing. Browsers throttle JavaScript timers in hidden tabs, so keep this tab visible during a trial.

Profile switching changes the answer mid-test. If a browser page or chat app grabs your microphone, many Bluetooth headsets drop from A2DP stereo audio into the hands-free (HFP) profile, and output latency and quality both change. Close the input app before measuring.

Toggle one audio processing feature at a time. ANC, “spatial audio”, Windows audio enhancements, and macOS Voice Isolation each add their own block delay, typically 5–15 ms apiece.

Two runs disagree wildly? Re-pair the headphones, move away from a congested Wi‑Fi channel, and check battery-saving modes, which throttle radio time-slots.

This tool times output only. Record-and-playback round trips, including the microphone’s analog-to-digital path and echo cancellation (often 10–40 ms on top), are best measured with the loopback monitoring page.

Related tools

• AV Sync Test — Check audio-versus-video drift with a flashing-and-clicking pattern instead of taps.
• Bluetooth Audio Latency Explained — Full codec-by-codec breakdown of why wireless delay happens and how to shrink it.
• Microphone Test — Verify the input side of the chain before diagnosing round-trip lag.
• Headphone Test — Confirm the headphones themselves produce clean output across the spectrum.