Use a Bluetooth headset's microphone without wrecking the audio quality
Checked on this Omarchy 4 workstation (omarchy 4.0.2-1, bluez 5.87-2, pipewire 1:1.6.8-1, wireplumber 0.5.15-1, kernel 7.1.9) and against the two cited Arch wiki pages, both refetched. The core of the record holds: bluetooth.autoswitch-to-headset-profile is still the setting name, still defaults to true, and is still read by WirePlumber 0.5.15. Confirmed on this machine in the schema block of /usr/share/wireplumber/wireplumber.conf and at lines 590 and 625 of /usr/share/wireplumber/scripts/device/autoswitch-bluetooth-profile.lua. wpctl settings --save KEY VAL exists in wpctl 0.5.15 (man wpctl), and ~/.config/wireplumber/wireplumber.conf.d/*.conf is the correct 0.5 location, matching the shipped template /usr/share/doc/wireplumber/examples/wireplumber.conf.d/bluetooth.conf. Four things were wrong for Omarchy 4. First, the mSBC advice is PulseAudio-era framing: man 7 pipewire-props on pipewire 1.6.8 defines bluez5.enable-msbc as "Override device quirk list and enable MSBC for devices for which it is disabled", so mSBC is already on except for quirked hardware and the record told the reader to enable something already enabled. Second, the record ignores that Omarchy 4 already installs a file into the exact directory it tells the reader to create: /usr/share/omarchy/config/wireplumber/wireplumber.conf.d/bluetooth-a2dp-autoconnect.conf, byte-identical to the copy present at ~/.config/wireplumber/wireplumber.conf.d/bluetooth-a2dp-autoconnect.conf on this machine, and omarchy refresh restores Omarchy's copy from $OMARCHY_PATH/config/. Third, the symptom names pavucontrol, which pacman -Q pavucontrol reports is not installed on Omarchy 4. pactl is present via libpulse 17.0+r98+gb096704c0-1. Fourth, the danger understated the cost of the roles override: man 7 pipewire-props gives the default role list as [ a2dp_sink a2dp_source bap_sink bap_source bap_bcast_sink bap_bcast_source hfp_hf hfp_ag ], so restricting it to the two A2DP roles also kills LE Audio, which the record did not say. The cause was also overstated as an absolute, since pipewire 1.6.8 ships duplex A2DP codecs and BAP, and the autoswitch script's own header comment names Faststream as a non-HFP mic path. Added an Omarchy branch for the restart step using omarchy-restart-audio, which also restarts pipewire-pulse (the record restarted only pipewire and wireplumber). Corrected the verify command: the original grep -A5 'Active Profile' prints the five lines after the profile line, which do not identify the card. NOT exercised: no Bluetooth device was paired, connected or profile-switched, because this is the operator's daily workstation, so the pactl set-card-profile commands and the codec-suffixed profile names were read from the plugin strings in /usr/lib/spa-0.2/bluez5/libspa-bluez5.so (headset-head-unit plus a %s-%s format) and from the shipped codec plugins libspa-codec-bluez5-hfp-msbc.so, libspa-codec-bluez5-hfp-lc3-swb.so and libspa-codec-bluez5-hfp-cvsd.so, not observed live.
bluez5.roles = [ a2dp_sink a2dp_source ] removes more than the headset microphone. On pipewire 1.6.8 the default role list is [ a2dp_sink a2dp_source bap_sink bap_source bap_bcast_sink bap_bcast_source hfp_hf hfp_ag ], so that line also disables LE Audio. A device that only speaks LE Audio will then produce no sink at all. Do not apply it if you take calls on those headphones, or if you own an LE Audio device.
Symptom
Bluetooth headphones sound good for music but the built-in mic is not offered as an input. Or the moment you join a call the output collapses to tinny mono and stays that way. pactl list cards | grep -E 'Name:|Active Profile:' shows the bluez_card.* entry flipping between a2dp-sink and headset-head-unit (or a codec-suffixed variant such as headset-head-unit-msbc), and wpctl status shows the stereo sink replaced by a low rate mono one. On Omarchy 4 there is no pavucontrol to watch this in, so read it from pactl or wpctl.
Cause
Classic Bluetooth cannot carry high quality stereo output and microphone input over A2DP at the same time. A2DP sink is output only, so using the mic means switching the whole card to the mono HFP profile, and the playback quality drops with it. WirePlumber performs that switch on its own: the setting bluetooth.autoswitch-to-headset-profile defaults to true in the schema in /usr/share/wireplumber/wireplumber.conf, and /usr/share/wireplumber/scripts/device/autoswitch-bluetooth-profile.lua switches whenever a client links to the device's loopback source node, which includes apps that only probe for a mic. The trade is no longer absolute. PipeWire 1.6.8 also carries duplex A2DP codecs (faststream_duplex, aptx_ll_duplex, opus_05_duplex) and LE Audio BAP, either of which can give a usable mic and good output together when both ends support them. Most common headsets still do not, so on those the trade is real and the switch is what you see.
Fix
Two settings decide this, and both live in WirePlumber 0.5. PulseAudio-era module-bluetooth-policy advice is dead on PipeWire.
Stop apps dragging the headset into HFP. You give up the headset mic while this is off:
wpctl settings --save bluetooth.autoswitch-to-headset-profile false
That key is current on wireplumber 0.5.15. Its schema is in /usr/share/wireplumber/wireplumber.conf under wireplumber.settings.schema with default = true, and /usr/share/wireplumber/scripts/device/autoswitch-bluetooth-profile.lua reads it through Settings.get_boolean.
To set it from a file instead, and to drop the HFP role entirely:
mkdir -p ~/.config/wireplumber/wireplumber.conf.d
# ~/.config/wireplumber/wireplumber.conf.d/51-mitigate-annoying-profile-switch.conf
wireplumber.settings = {
bluetooth.autoswitch-to-headset-profile = false
}
monitor.bluez.properties = {
bluez5.roles = [ a2dp_sink a2dp_source ]
}
**On Omarchy 4 that directory is not empty.** Omarchy ships config/wireplumber/wireplumber.conf.d/bluetooth-a2dp-autoconnect.conf and copies it into ~/.config/wireplumber/wireplumber.conf.d/. It sets bluez5.auto-connect = [ a2dp_sink a2dp_source ] through a monitor.bluez.rules entry. Add your own file next to it rather than editing it, because omarchy refresh restores Omarchy's copy from $OMARCHY_PATH/config/.
If you want the headset mic to work well instead, leave auto-switching on and **do not reach for bluez5.enable-msbc**. On pipewire 1.6.8 mSBC is already active for every device the quirk database does not exclude. man 7 pipewire-props describes the property as "Override device quirk list and enable MSBC for devices for which it is disabled", so it is an override, not a switch. Set it only when your specific headset is on the quirk list:
# ~/.config/wireplumber/wireplumber.conf.d/90-bluez-force-codecs.conf
monitor.bluez.properties = {
bluez5.enable-msbc = true
bluez5.enable-sbc-xq = true
}
Restart audio, then reconnect the device. On Omarchy 4 use the wrapper, which also restarts pipewire-pulse and recovers a stuck USB audio device:
omarchy-restart-audio
On plain Arch:
systemctl --user restart wireplumber.service pipewire.service pipewire-pulse.service
Switch profile by hand at any time. pactl is present on Omarchy 4 through libpulse, but pavucontrol is not installed:
pactl list cards short
pactl set-card-profile bluez_card.XX_XX_XX_XX_XX_XX a2dp-sink
The HFP profile name carries the negotiated codec, built as headset-head-unit-<codec>, so it may be headset-head-unit-cvsd, headset-head-unit-msbc or headset-head-unit-lc3_swb depending on what the headset agreed to. List what your card actually offers before naming one:
pactl list cards | grep -E 'headset-head-unit|a2dp-sink'
Verify
wpctl settings bluetooth.autoswitch-to-headset-profile
prints false once saved (a bare wpctl settings KEY only reads it). Then:
pactl list cards | grep -E 'Name:|Active Profile:'
shows Active Profile: a2dp-sink for the bluez_card.* entry and it stays there when you open a video call site. To see which HFP profiles the headset actually offers, rather than assuming mSBC:
pactl list cards | grep -E 'headset-head-unit'