Why Bluetooth Sound Drops When the Mic Turns On
Because the link stops being a music link. The moment something asks a Bluetooth headset for its microphone, the connection switches from A2DP, the stereo playback profile, to HFP, the mono headset profile, and HFP was built to carry speech on a phone call, not sound. That is why Bluetooth audio quality drops when the mic is on, and iOS gives you no switch to stop it.
The switch only happens when a Bluetooth device is asked to be the input. A setup that takes the microphone from the phone and sends audio out over Bluetooth never triggers it. You can usually see the collapse reflected in the numbers a browser microphone test reports, and hear it in a five second recording either way, and the rest of this page explains what both are telling you.
One Bluetooth link, one audio profile at a time
A Bluetooth connection is not a single pipe. It is a set of profiles, and audio uses two very different ones.
A2DP (Advanced Audio Distribution Profile) runs one direction only, phone to speaker. It is stereo, it gets a generous slice of the radio's bandwidth, and on an iPhone it carries AAC, with SBC as the fallback. This is what plays your music.
HFP (Hands-Free Profile) has to carry audio both ways at once so you can talk and listen. To do that in real time it opens a synchronous voice channel, and that channel has a tiny budget. Mono, low sample rate, speech codec.
The two audio streams do not run together on the same link. When the voice channel comes up, the A2DP stream is suspended, and when it drops, A2DP resumes after a short gap while the codec re-syncs. Most people recognise that gap before they can name it: music stops, a soft click, and everything comes back sounding like a drive-through intercom.
Why Bluetooth audio quality drops when the mic is on: a2dp vs hfp
| A2DP (music profile) | HFP (headset profile) | |
|---|---|---|
| Direction | Phone to speaker only | Both ways at once |
| Channels | Stereo | Mono |
| Codec on iPhone | AAC, SBC as fallback | CVSD narrowband, or mSBC where both ends support Wideband Speech |
| Sample rate | 44.1 or 48 kHz | 8 kHz narrowband, 16 kHz wideband |
| Usable audio bandwidth | Full range | Roughly 4 kHz, or 8 kHz on wideband |
| Sounds like | Music | A phone call |
An 8 kHz sample rate means nothing above about 4 kHz survives. That is not a subtle loss. Cymbals disappear, sibilance turns to mush, and the consonants that carry intelligibility in speech, the s and f and th sounds, live right in the band that gets thrown away. This is the whole reason a Bluetooth headset mic sounds bad on a recording even though the same headset sounds fine playing back music.
If your Bluetooth audio quality drops on a call and comes back the second you hang up, nothing is broken. You are hearing the profile switch, and it is behaving exactly as specified.
Hear the switch happen, and watch the number fall
You do not have to take this on trust. Connect a Bluetooth headset with a microphone, play something through it, then open the microphone test in this tab and allow microphone access. The first change is the one you hear: the music quality collapses. The second is on screen, where the tool reports the input sample rate and channel count as the browser exposes them. On a headset that has switched to the headset profile that readout usually falls to narrowband mono, typically 16000 or 8000 and one channel instead of 48000 and two. Browsers differ in which of those values they populate, and some resample before reporting, so read the numbers as a strong hint rather than proof.
That caveat matters more than it sounds. A browser that resamples will happily keep reporting 48 kHz while the link is running narrowband underneath, and a browser that leaves the channel count unset tells you nothing either way. When the numbers look unchanged but the sound has clearly degraded, trust your ears and use the five second record-and-play-back in the same tool instead. The playback settles it, because you cannot resample back the frequencies that were never captured.
Run the test twice: once with the headset connected, once with it off and the phone or laptop using its own microphone. Same voice, same room, same tool. The second recording is what a microphone is supposed to sound like.
iOS gives you no control over the switch
On a Mac you can set the input device and the output device separately in Sound settings, which lets you keep A2DP playback while recording from something else. iOS has no equivalent. There is no per-app input picker, no toggle in Settings, and no public API that lets an app keep a headset on the music profile while borrowing its microphone.
That last part is not iOS being stubborn. A2DP has no return path at all. There is no microphone channel in the profile to keep open, so there is nothing for iOS to expose. Any app that wants a Bluetooth device's microphone has to ask for the headset profile, and asking is what costs you the music.
What an app can do is decline to ask. An audio session that only requests Bluetooth for playback leaves the link on A2DP, because nothing ever requests an input from the speaker.
The setup that sidesteps the switch entirely
Take the microphone from the phone and send the audio out over Bluetooth. The phone's own mic array is the input, the Bluetooth speaker is only ever an output, and the link therefore stays on A2DP for the whole session. No profile renegotiation, no narrowband, no mono.
That is the architecture behind the free iPhone and iPad app this site is built around, and it is the reason it sounds better than talking into a Bluetooth headset. You pair the speaker in iOS Settings, which is the only place any iPhone can do that, then the app routes the live microphone signal to whatever output iOS has already chosen. There is a browser version of the same chain on the live mic monitor page if you want to hear the passthrough before installing anything. The full walkthrough is in the guide on connecting an iPhone microphone to a Bluetooth speaker.
One exception worth knowing. If the output device has its own microphone, such as AirPods or a headset, iOS may still choose that microphone and drop the link into headset mode. A plain Bluetooth speaker with no microphone has nothing to switch to, so it stays on the music profile. If your sound suddenly goes narrowband, check which input the system picked using the microphone test.
Avoiding the switch does not fix the delay
Staying on A2DP protects the sound. It does nothing about latency, and no app can. Hollyland puts A2DP end to end at roughly 100 to 300 ms. AAC, the codec an iPhone actually uses, is quoted at roughly 100 to 200 ms. SBC is quoted as both 100 to 150 ms and 150 to 300 ms depending on which source you read, and those sources genuinely disagree, so we present the range rather than averaging it into a number nobody published.
Qualcomm's aptX Low Latency sits at about 40 ms, which sounds appealing until you learn that iPhone has no aptX support at all. LC3 on LE Audio is quoted at 20 to 30 ms by the Bluetooth SIG. Dedicated 2.4 GHz wireless microphone systems come in under 20 ms, which is why professionals buy them.
We are not going to quote a millisecond figure for the headset profile, because none of the sources we trust publish one and we have not measured it. What we can tell you is your own number: the Bluetooth latency test plays a chirp, records it back through your microphone, and cross-correlates five runs to report the real round trip on your gear. For the sourced comparison, see the guide on SBC, AAC and aptX latency, the wider picture in Bluetooth microphone delay, and the gap a cable closes in wired versus Bluetooth microphones.
When you actually want the headset profile
Sometimes the switch is correct. On a phone call, a FaceTime call, a voice memo dictated while driving, or a Siri request, the microphone is the whole point and mono narrowband speech is a fair price for keeping your hands free.
The switch is only a problem when you wanted both at once: good sound out and a microphone in. For that, split the jobs. Microphone on the phone, audio out over Bluetooth, or a cable if the room is unforgiving. If you are amplifying a voice into a room, also read what happens when the speaker feeds back into the microphone in the guide on microphone feedback with a Bluetooth speaker, and why Live Listen will not route to a Bluetooth speaker.
A short checklist before it matters
- Decide which device is the microphone before you start. If it is a Bluetooth headset, accept narrowband mono and move on.
- If sound quality matters, use the phone's own microphone and keep the Bluetooth link for playback only.
- Confirm the input with the microphone test. A steady 48000 and two channels points to the music profile still being in charge; if the browser reports nothing useful, record five seconds and listen back.
- Measure your round trip once on the gear you will actually use, so the delay is a known quantity rather than a surprise.
- Keep a wired option in the bag for anything where timing has to be tight.
The profile switch is one of the few Bluetooth behaviours that is fully explainable and fully avoidable, as long as you never ask a speaker to be a microphone. The rest of all the guides works through the parts that are harder to dodge, and the home page has the app and the browser tools in one place.
Test the microphone before you rely on it
Check the level, the sample rate and the channel count of any microphone — including the moment a Bluetooth headset drops to narrowband.