[pulseaudio-discuss] [BUG] Using bluez 5.5 & pulseaudio a bluetooth headset cannot be used as a recording device.

[Tanu Kaskinen]

On Fri, 2013-05-31 at 07:14 +0200, Alexander Winnig wrote:
> Am 30.05.2013 15:05, schrieb Tanu Kaskinen:
> > [Added pulseaudio-discuss to CC again.]
> >
> > On Thu, 2013-05-30 at 14:39 +0200, Alexander Winnig wrote:
> >>> Idea: use BlueZ 4, and don't try to use the bluetooth alsa plugin.
> >>> PulseAudio has native support for BlueZ, so there's no need to have alsa
> >>> in between.
> >> I did, arecord and aplay still don't work with it.
> > Does parecord work any better?
> >
> See here ( http://www.youtube.com/watch?v=E-kWdpIc7ig ) what I did so 
> far. I recommend the big player and 480p.

Some notes about the video:

At 1:21 the available sinks and sources only include the auto_null sink
and its monitor source. Why isn't the RPi's own sound card getting
detected? Well, let's not worry too much about that

At 2:13 we can see that BlueZ has a device for the headset. None of the
audio profiles are connected, so the device isn't visible as an audio
device in PulseAudio.

At 2:22 HSP profile connection is attempted, but it fails.

At 2:41 headset emulation is enabled, which I guess means the HFGW
profile, so you make RPi appear as a headset to other Bluetooth devices.
This is not relevant for your use case, although I don't think it should
do any harm either. bluetoothd gets restarted at this point.

At 2:50 HSP profile connection is attempted again, and it fails again.

At 3:04 bluetoothd might get restarted again, depending on whether
clicking "Apply" actually did something or not (no settings were
changed).

At 3:24 the headset device is removed from BlueZ.

At 3:56 the headset is paired again. The HSP profile is connected at the
same time, this time successfully. This should make the headset visible
as an audio device in PulseAudio.

At 4:23 we see that there still isn't other sinks than auto_null. Either
the bluetooth card profile is "off", or PulseAudio had problems with
creating a sink for the headset.

At 4:56 arecord is used to access the headset through the bluetooth alsa
plugin. It doesn't seem to work (test.wav doesn't get created).

At 6:58 parecord is used to record from an unspecified source. Given
that the only source was auto_null.monitor last time the source list was
checked, I'm pretty sure that parecord records from auto_null.monitor.

At 7:22 aplay is used to access the headset through the bluetooth alsa
plugin. It doesn't seem to work.

At 9:05 pavucontrol is started. It lists both the RPi sound card and the
headset as output devices. This is clearly not in line with the previous
"pactl list sinks" command. The headset is also visible in the input
device list.

Conclusion: remove the .asoundrc file. It's totally unnecessary, and the
bluetooth alsa plugin might interfere with PulseAudio's ability to
access the headset. This is not the biggest problem, though - the real
blocker is that on the command line pactl and friends access a different
server than pavucontrol. I don't know the reason for that. The
"pulseaudio -k" command in the beginning doesn't find any running
daemons, although I suspect that there is a daemon running.

Have you compiled pulseaudio from source? Have you edited .bashrc to set
up e.g. LD_LIBRARY_PATH, resulting in a different environment in the
shell compared to the LXDE desktop environment, from which pavucontrol
is launched?

-- 
Tanu