Dictation cuts off the first words when you use AirPods on a Mac (Wispr Flow, Superwhisper, VoiceInk)

By Hardeep Singh, maker of CallLane · Updated

The AirPods mic is off until you start dictating, and turning it on switches the AirPods into headset mode, which takes a moment, so the first words are lost. Dictate through a mic that doesn't need that switch, your Mac's built-in mic or a USB mic, and keep the AirPods for listening. Pick it by name in the dictation app's microphone setting. If the app stays on the system default, CallLane can keep the system input on the Mac's mic; it doesn't remove any start-up delay in the dictation app itself.

Written from Apple’s and the app makers’ own documentation and the CallLane 1.0.5 source, . Not yet tested with each dictation app, and the start delay hasn't been measured.

CallLane keeps your music at full volume on calls and your AirPods in high quality. One-time purchase, 14-day refund.

Buy $9

Why the first words go missing

While you’re only listening, AirPods use a stereo listening mode with no mic. When a dictation app starts recording from them, the Mac and the AirPods have to set up a second Bluetooth link for the mic and switch the sound to headset mode. Until that link is up, there is no audio to record, and whatever you said in that gap is gone.

The same complaint shows up across dictation apps:

  • A Superwhisper user on Reddit lost 1 to 3 seconds each time; a Superwhisper team member replied that Bluetooth is tricky, that some apps keep the mic open all the time to avoid it, and that wired setups typically don’t have the issue (thread).
  • In VoiceInk, push-to-talk starts about half a second late, longer after the Mac has been idle, with a Bose QC45 (issue #996). After moving the input off the headset with CallLane, the reporter said on 25 October 2026 that this was only part of the delay: about a second still remained.
  • In MacParakeet, the first recording with AirPods Pro fails and the second works (issue #1238).

There’s a second cost. In headset mode, everything you hear is mono call-quality audio until the recording stops.

In headset mode, Bluetooth carries the sound both ways through a speech codec: CVSD at 8 kHz or mSBC at 16 kHz on most headphones. On AirPods 3 and the original AirPods Pro, a developer who read macOS’s logs in February 2022 found AAC-ELD at 24 kHz (reported here). On AirPods Pro 3 with macOS 27.0.1, I measured their sound at 24 kHz mono while the mic was open (the measurement); macOS doesn’t say which codec that is. All of these are mono and far below the stereo of the listening mode.

The fix: dictate through the Mac’s mic

The built-in mic, or a USB mic, needs no Bluetooth switch, so the profile-switch part of the delay goes away. Any delay the dictation app has in starting its own recording is separate, and this doesn’t fix it. Your AirPods stay the output, in full quality. Wispr’s own help page gives the same advice: it warns that AirPods “may be slow to start and have more errors in dictation” and recommends the built-in microphone.

Some apps now do this on their own. MacParakeet switched to the built-in mic by default when the output is Bluetooth (pull request #613).

What about keeping the mic warm?

The other fix is to keep the mic open between dictations, so it’s ready when you speak. EnviousWispr holds it open for 30 seconds by default, and macos-mic-keepwarm keeps it open for other apps. If the warm mic is the AirPods mic, the AirPods stay in headset mode the whole time, so your music stays muffled. It trades one problem for the other. Keeping the Mac’s mic warm avoids the headset-mode cost, though I haven’t measured how much start delay it removes.

Where CallLane fits

You don’t need CallLane for this; picking the Mac’s mic in your dictation app is the fix. CallLane helps when the app is on Auto-detect or System default: its Lock input holds the Mac’s system input on the mic you chose, so macOS can’t move it to the AirPods on reconnect.

One limit. macOS can make the AirPods the input at the moment an app starts its mic. CallLane switches it back as soon as macOS reports the change, but an app that grabs the mic in that instant can still get the AirPods. Picking the built-in mic by name in the dictation app avoids that. I haven’t measured the start delay with CallLane on, so I won’t put a number on it.

Questions

Why does Wispr Flow or Superwhisper miss my first words with AirPods?

Turning on the AirPods mic switches them to headset mode, which takes a moment, and nothing is recorded until the switch is done. Dictating through the Mac's built-in mic removes that switch. Any start-up delay in the dictation app itself is separate.

Does keeping the mic warm fix the delay?

It can remove the switch delay, but if the warm mic is the AirPods mic, they stay in low-quality headset mode while it's open. Using the Mac's mic avoids the headset-mode switch without that cost.

CallLane · $9One-time · 3 Macs · 14-day refund
Buy