Wired · Bluetooth · speakers
Audio latency test
How late does the sound reach your ears? Tap along to a click, or let the mic time it, for wired headphones, Bluetooth earbuds or speakers.
How to use it
- Pick what you listen on, Wired, Bluetooth or Speakers, and tap Start tapping. Tap the pad, or press the space bar, on each click you hear, as steadily as you can, until it shows your delay.
- For a more exact number, open More. With wired headphones or speakers, hold an earbud or ear cup near the mic and tap Start listening: it plays a few clicks and times them. With Bluetooth earbuds, keep them in and clap along to 12 clicks. Each takes about 10 seconds; test another pair to see them side by side.
What your number means
The result puts your delay on a bar from 0 to 300 ms, in four zones, each with what it means for singing. The thresholds are a guide, not a law: some people notice a little less, some a little more.
| Zone | Delay | For singing and recording |
|---|---|---|
| Tight | Under 10 ms | Most people won't notice it; hearing yourself live, some singers still will. |
| Fine | 10 to 39 ms | Fine for recording; noticeably late to hear yourself live. |
| Late | 40 to 199 ms | Fine for stacking harmonies in an app that lines takes up; too slow to hear yourself live. |
| Very late | 200 ms and up | Singing along will drag. Use wired headphones if you can. |
Bluetooth latency: typical delays by connection
Rough ranges, as they're often measured. Your own headphones, phone and browser can sit well outside them, which is the reason to test. Phones and computers also add their own buffering on top of the headphones.
Your Bluetooth codec is chosen by the two devices between them. An iPhone uses AAC when the headphones support it, or SBC when they don't, and never aptX; many Android phones show the codec in use under Developer options.
| Connection | Often | Why |
|---|---|---|
| Wired headphones | Under 50 ms | Only the device's own buffering: the cable adds nothing you could hear. |
| Built-in speakers | Under 50 ms, plus the air | The same buffering, plus about 3 ms for every metre between you and the speaker. |
| Bluetooth (SBC, AAC) | 100 to 300 ms, sometimes more | Compressing, sending and buffering the sound on the headphones. |
| Bluetooth low-latency (aptX Low Latency, aptX Adaptive, LE Audio, game modes) | About 40 to 150 ms | Smaller buffers and a faster codec, when both ends support it. |
Two ways to measure
Tap along: a steady click plays, 100 a minute, and you tap the pad or the space bar on each one. The test keeps the middle of your taps, drops the ones far off, and shows the spread as a plus-or-minus. It works anywhere, with no mic, but your own timing is part of the number.
Let the mic hear it: under More, with wired headphones or speakers, it plays a train of short clicks and listens for them through the mic, three times, and keeps the middle result. Hold an earbud or ear cup right by the mic, or with speakers turn the volume up in a quiet room. It measures the whole trip, out through the headphones and back in through the mic, which is exactly the delay a recording app has to correct for.
With Bluetooth earbuds, which stop playing once they're out of your ear, you keep them in and clap along to 12 clicks instead, or tap the table. The phone's or laptop's own mic hears your claps, and since people clap a little ahead of the beat, that usual lead is added back.
Singing over a track with Bluetooth
When you sing along to earlier parts, you hear them late, and you sing in time with what you hear. Then your voice takes a little longer again to reach the recording. So a take made over Bluetooth comes out late by the whole round trip, which sounds like a sloppy entry on every note.
The fix is to measure that delay once and move each take back by it. Stackapella does this for you: it starts from a typical Bluetooth delay, and a short check where you clap along to clicks, earbuds in, measures yours, so every part you record lines up with the parts you heard. What no app can fix is hearing yourself live through Bluetooth: for that, use wired headphones.
Free tools by Stackapella
Your delay's measured. The app lines your parts up for it.
Stackapella teaches you to sing harmony one part at a time, right in your browser.
- A guide sings your partHear your line over the others, slow it down, then take your turn.
- Your notes light upEach note shows where you landed, and a hint when to go up or down.
- Record every part yourselfStack your takes in time until you're the whole choir.
- Every course free
- Open source
- No ads, no tracking
Questions
How much audio latency is noticeable?
It depends on what you're doing. Under about 25 ms most people won't notice it as a delay. Hearing your own voice live in headphones is stricter: some singers notice as little as 5 to 10 ms, more as a change in tone than an echo. Past that, your own voice starts to sound late in your ears, though a steady delay is still fine for recording over a track in an app that lines your takes up.
Why is Bluetooth audio delayed?
Because the sound is squeezed into small packets, sent over radio, held for a moment on the headphones in case a packet goes missing, then unpacked again. Each step takes time. With everyday Bluetooth headphones the total is often 100 to 300 ms, sometimes more, far more than a cable.
Do wired headphones have less latency than Bluetooth?
Yes, much less. The cable itself adds nothing you could hear, so what's left is the device's own audio buffering, the same delay you'd get from its speakers. Test both here and the difference is the Bluetooth part.
How much delay is OK for singing along to a track?
Any steady delay is fine for recording, as long as the app lines your take up afterwards. Hearing yourself live is different: that needs very little delay, about 10 ms or less for many singers, which in practice means wired headphones. Bluetooth is too slow to hear your own voice through without an echo.
Can I record with Bluetooth earbuds?
Yes, to listen, but sing into the phone or laptop mic, not the earbuds' own. On most Bluetooth headphones, using their mic switches them to call mode, which makes everything you hear sound like a phone call and changes the delay. The test warns you about this when you pick Bluetooth.
How do I reduce Bluetooth latency?
Use wired headphones if you can. If not, turn on a game or low-latency mode if your headphones have one, and use a low-latency codec, such as aptX Low Latency, aptX Adaptive or LE Audio, if both your device and headphones support it. Test again after each change to see what it did.
How accurate is a tap test, and why is the mic check better?
A tap test is a good rough guide: your own timing goes into the number, since nobody taps exactly on the beat. That's why it asks for 8 taps, drops the uneven ones and shows a plus-or-minus, and asks for more while that spread is wide. The mic check takes your hands out of it: with wired headphones or speakers it times the clicks themselves, out through the headphones and back in through the mic, three times over, so its number runs a little above what your ears alone hear. Bluetooth earbuds stop playing once they're out of your ear, so with them you keep them in and clap along to 12 clicks instead: the mic hears your claps, and the usual lead of a clap on the beat is added back.
Why does my browser report a different number?
Your browser only knows its own part of the delay, and often not what the Bluetooth link adds. After a test, the result shows what your browser says beside what was measured, so you can see the gap.
Is this the same as internet latency or ping?
No. Ping is how long data takes to reach a server and come back over the internet. This test measures the delay inside your own device and headphones, and works with no connection at all.
Is my voice or anything else recorded?
No. Tapping never uses the mic. The mic opens only when you start a mic check (Start listening, or Start the clicks with Bluetooth), to hear the clicks or your claps, and nothing is recorded, kept or sent.
More free tools
No account, no ads. Each one keeps working offline once you've opened it.
See all 17 tools