This is a pre-production deployment of Warehouse. Changes made here affect the production instance of PyPI (
Help us improve Python packaging - Donate today!

Python high-level interface and ctypes-based bindings for PulseAudio (libpulse)

Project Description
python-pulse-control (pulsectl module)

Python (3.x and 2.x) high-level interface and ctypes-based bindings for
PulseAudio_ (libpulse), mostly focused on mixer-like controls and
introspection-related operations (as opposed to e.g. submitting sound samples to
play, player-like client).

Originally forked from pulsemixer_ project, which had this code bundled.

.. _PulseAudio:
.. _pulsemixer:


.. contents::
:backlinks: none


Simple example::

from pulsectl import Pulse

with Pulse('volume-increaser') as pulse:
for sink in pulse.sink_list():
# Volume is usually in 0-1.0 range, with >1.0 being soft-boosted
pulse.volume_change_all_chans(sink, 0.1)

Listening for server state change events::

from pulsectl import Pulse, PulseLoopStop

with Pulse('event-printer') as pulse:
# print('Event types:', ', '.join(pulse.event_types))
# print('Event facilities:', ', '.join(pulse.event_facilities))
# print('Event masks:', ', '.join(pulse.event_masks))

def print_events(ev):
print('Pulse event:', ev)
### Raise PulseLoopStop for event_listen() to return before timeout (if any)
# raise PulseLoopStop


Misc other tinkering::

>>> from pulsectl import Pulse
>>> pulse = Pulse('my-client-name')

>>> pulse.sink_list()
[<PulseSinkInfo at 7f85cfd053d0 - desc='Built-in Audio', index=0L, mute=0, name='alsa-speakers', channels=2, volumes='44.0%, 44.0%'>]

>>> pulse.sink_input_list()
[<PulseSinkInputInfo at 7fa06562d3d0 - index=181L, mute=0, name='mpv Media Player', channels=2, volumes='25.0%, 25.0%'>]

>>> pulse.sink_input_list()[0].proplist
{'application.icon_name': 'mpv',
'application.language': 'C',
'': 'mpv Media Player',
'native-protocol.version': '30',
'window.x11.display': ':1.0'}

>>> pulse.source_list()
[<PulseSourceInfo at 7fcb0615d8d0 - desc='Monitor of Built-in Audio', index=0L, mute=0, name='alsa-speakers.monitor', channels=2, volumes='100.0%, 100.0%'>,
<PulseSourceInfo at 7fcb0615da10 - desc='Built-in Audio', index=1L, mute=0, name='alsa-mic', channels=2, volumes='100.0%, 100.0%'>]

>>> sink = pulse.sink_list()[0]
>>> pulse.volume_change_all_chans(sink, -0.1)
>>> pulse.volume_set_all_chans(sink, 0.5)

>>> pulse.server_info().default_sink_name
>>> pulse.default_set(sink)

>>> card = pulse.card_list()[0]
>>> card.profile_list
[<PulseCardProfileInfo at 7f02e7e88ac8 - description='Analog Stereo Input', n_sinks=0, n_sources=1, name='input:analog-stereo', priority=60>,
<PulseCardProfileInfo at 7f02e7e88b70 - description='Analog Stereo Output', n_sinks=1, n_sources=0, name='output:analog-stereo', priority=6000>,
<PulseCardProfileInfo at 7f02e7e9a4e0 - description='Off', n_sinks=0, n_sources=0, name='off', priority=0>]
>>> pulse.card_profile_set(card, 'output:hdmi-stereo')

>>> help(pulse)

>>> pulse.close()

Current code logic is that all methods are invoked through the Pulse instance,
and everything returned from these are "Pulse-Something-Info" objects - thin
wrappers around C structs that describe the thing, without any methods attached.

Pulse client can be integrated into existing eventloop (e.g. asyncio, twisted,
etc) using ``Pulse.set_poll_func()`` or ``Pulse.event_listen()`` in a separate

Somewhat extended usage example can be found in `pulseaudio-mixer-cli`_ project

.. _pulseaudio-mixer-cli:


Some less obvious things are described in this section.

Things not yet wrapped/exposed in python

There are plenty of information, methods and other things in libpulse not yet
wrapped/exposed by this module, as they weren't needed (yet) for author/devs

Making them accessible from python code can be as simple as adding an attribute
name to the "c_struct_fields" value in PulseSomethingInfo objects.

See `github #3 <>`_
for a more concrete example of finding/adding such stuff.

For info and commands that are not available through libpulse introspection API,
it is possible to use ``pulsectl.connect_to_cli()`` fallback function, which
will open unix socket to server's "module-cli" (signaling to load it, if
necessary), which can be used in exactly same way as "pacmd" tool (not to be
confused with "pactl", which uses native protocol instead of module-cli) or
pulseaudio startup files (e.g. "").

Probably a bad idea to parse string output from commands there though, as these
are not only subject to change, but can also vary depending on system locale.


In PulseAudio, "volume" for anything is not a flat number, but essentially a
list of numbers, one per channel (as in "left", "right", "front", "rear", etc),
which should correspond to channel map of the object it relates/is-applied to.

In this module, such lists are represented by PulseVolumeInfo objects.

I.e. ``sink.volume`` is a PulseVolumeInfo instance, and all thin/simple wrappers
that accept index of the object, expect such instance to be passed, e.g.
``pulse.sink_input_volume_set(sink.index, sink.volume)``.

There are convenience ``volume_get_all_chans``, ``volume_set_all_chans`` and
``volume_change_all_chans`` methods to get/set/adjust volume as/by a single
numeric value, which is also accessible on PulseVolumeInfo objects as a
``value_flat`` property.

PulseVolumeInfo can be constructed from a numeric volume value plus number of
channels, or a python list of per-channel numbers.

All per-channel volume values in PulseVolumeInfo (and flat values in the wrapper
funcs above), are float objects in 0-65536 range, with following meaning:

* 0.0 volume is "no sound" (corresponds to PA_VOLUME_MUTED).

* 1.0 value is "current sink volume level", 100% or PA_VOLUME_NORM.

* >1.0 and up to 65536.0 (PA_VOLUME_MAX / PA_VOLUME_NORM) - software-boosted
sound volume (higher values will negatively affect sound quality).

Probably a good idea to set volume only in 0-1.0 range and boost volume in
hardware without quality loss, e.g. by tweaking sink volume (which corresponds
to ALSA/hardware volume), if that option is available.

Note that ``flat-volumes=yes`` option ("yes" by default on some distros, "no" in
e.g. Arch Linux) in pulseaudio daemon.conf already scales device-volume with the
volume of the "loudest" application, so already does what's suggested above.

Fractional volume values used in the module get translated (in a linear fashion)
to/from pa_volume_t integers for libpulse. See ``src/pulse/volume.h`` in
pulseaudio sources for all the gory details on the latter (e.g. how it relates
to sound level in dB).

Code example::

with Pulse('volume-example') as pulse:
sink_input = pulse.sink_input_list()[0] # first random sink-input stream

volume = sink_input.volume
print(volume.values) # list of per-channel values (floats)
print(volume.value_flat) # average level across channels (float)


volume.value_flat = 0.3 # sets all volume.values to 0.3
pulse.volume_set(sink_input, volume) # applies the change


n_channels = len(volume.values)
new_volume = PulseVolumeInfo(0.5, n_channels) # 0.5 across all n_channels
# new_volume = PulseVolumeInfo([0.15, 0.25]) # from a list of channel levels (stereo)
pulse.volume_set(sink_input, new_volume)
# pulse.sink_input_volume_set(sink_input.index, new_volume) # same as above

In most common cases, doing something like
``pulse.volume_set_all_chans(sink_input, 0.2)`` should do the trick though -
no need to bother with specific channels in PulseVolumeInfo there.

String values

libpulse explicitly returns utf-8-encoded string values, which are always
decoded to "abstract string" type in both python-2 (where it's called "unicode")
and python-3 ("str"), for consistency.

It might be wise to avoid mixing these with encoded strings ("bytes") in the code,
especially in python-2, where "bytes" is often used as a default string type.

Enumerated/named values (enums)

In place of C integers that correspond to some enum or constant (e.g. -1 for
PA_SINK_INVALID_STATE), module returns EnumValue objects, which are comparable
to strings ("str" type in py2/py3).

For example::

>>> pulsectl.PulseEventTypeEnum.change == 'change'
>>> pulsectl.PulseEventTypeEnum.change
<EnumValue event-type change>
>>> pulsectl.PulseEventTypeEnum
<Enum event-type [change new remove]>

It might be preferrable to use enums instead of strings in the code so that
interpreter can signal error on any typos or unknown values specified, as
opposed to always silently failing checks with bogus strings.

Event-handling code, threads

libpulse clients always work as an event loop, though this module kinda hides
it, presenting a more conventional blocking interface.

So what happens on any call (e.g. ``pulse.mute(...)``) is:

* Make a call to libpulse, specifying callback for when operation will be completed.
* Run libpulse event loop until that callback gets called.
* Return result passed to that callback call, if any (for various "get" methods).

``event_callback_set()`` and ``event_listen()`` calls essentally do raw first
and second step here.

Which means that any pulse calls from callback function can't be used when
``event_listen()`` (or any other pulse call through this module, for that
matter) waits for return value and runs libpulse loop already.

One can raise PulseLoopStop exception there to make ``event_listen()`` return,
run whatever pulse calls after that, then re-start the ``event_listen()`` thing.

This will not miss any events, as all blocking calls do same thing as
``event_listen()`` does (second step above), and can cause callable passed to
``event_callback_set()`` to be called (when loop is running).

Also, same instance of libpulse eventloop can't be run from different threads,
naturally, so if threads are used, client can be initialized with
``threading_lock=True`` option (can also accept lock instance instead of True)
to create a mutex around step-2 (run event loop) from the list above, so
multiple threads won't do it at the same time.

For proper eventloop integration (think twisted or asyncio), ``_pulse_get_list``
/ ``_pulse_method_call`` wrappers should be overidden to not run pulse loop, but
rather return "future" object and register a set of fd's (as passed to
``set_poll_func`` callback) with eventloop.
Never needed that, so not implemented in the module, but should be rather easy
to implement on top of it, as described.


Test code is packaged/installed with the module and can be useful to run when
changing module code, or to check if current python, module and pulseudio
versions all work fine together.

Commands to run tests from either checkout directory or installed module::

% python2 -m unittest pulsectl.tests.all
% python3 -m unittest pulsectl.tests.all

Note that if "pulsectl" module is available both in current directory
(e.g. checkout dir) and user/system python module path, former should always
take priority for commands above.

Test suite runs ad-hoc isolated pulseaudio instance with null-sinks (not
touching hardware), custom (non-default) startup script and environment,
and interacts only with that instance, terminating it afterwards.
Still uses system/user daemon.conf files though, so these can affect the tests.

Any test failures can indicate incompatibilities, bugs in the module code,
issues with pulseaudio (or its daemon.conf) and underlying dependencies.
There are no "expected" test case failures.

Changelog and versioning scheme

This package uses one-version-per commit scheme (updated by pre-commit hook)
and pretty much one release per git commit, unless more immediate follow-up
commits are planned or too lazy to run ``py sdist bdist_wheel upload``
for some trivial README typo fix.

| Version scheme: ``{year}.{month}.{git-commit-count-this-month}``
| I.e. "16.9.10" is "11th commit on Sep 2016".
There is a `CHANGES.rst <CHANGES.rst>`_ file with the list of any intentional
breaking changes (should be exceptionally rare, if any) and new/added non-trivial

| It can be a bit out of date though, as one has to remember to update it manually.
| "Last synced/updated:" line there might give a hint as to by how much.


It's a regular package for Python (3.x or 2.x).

Be sure to use python3/python2, pip3/pip2, easy_install-... binaries below,
based on which python version you want to install the module for, if you have
several on the system (as is norm these days for py2-py3 transition).

Using pip_ is the best way::

% pip install pulsectl

(add --user option to install into $HOME for current user only)

Or, if you don't have "pip" command::

% python -m ensurepip
% python -m pip install --upgrade pip
% python -m pip install pulsectl

(same suggestion wrt "install --user" as above)

On a very old systems, one of these might work::

% curl | python
% pip install pulsectl

% easy_install pulsectl

% git clone --depth=1
% cd python-pulse-control
% python install

(all of install-commands here also have --user option)

Current-git version can be installed like this::

% pip install 'git+'

Note that to install stuff to system-wide PATH and site-packages (without
--user), elevated privileges (i.e. root and su/sudo) are often required.

Use "...install --user", `~/.pydistutils.cfg`_ or virtualenv_ to do unprivileged
installs into custom paths.

More info on python packaging can be found at ``_.

.. _pip:
.. _~/.pydistutils.cfg:
.. _virtualenv:


* pulsemixer_ - initial source for this project (embedded in the tool).

* `libpulseaudio <>`_ -
different libpulse bindings module, more low-level, auto-generated from
pulseaudio header files.

Branches there have bindings for different (newer) pulseaudio versions.

* `pypulseaudio <>`_ -
high-level bindings module, rather similar to this one.

* `pulseaudio-mixer-cli`_ - alsamixer-like script built on top of this module.
Release History

Release History

This version
History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


History Node


Download Files

Download Files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
pulsectl-17.6.1-py2.py3-none-any.whl (33.9 kB) Copy SHA256 Checksum SHA256 2.7 Wheel Jun 20, 2017
pulsectl-17.6.1.tar.gz (32.5 kB) Copy SHA256 Checksum SHA256 Source Jun 20, 2017

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting