Logo

HomeAboutCommunityDownloadDocumentationPlanet


PulseAudio 15.0 release notes (draft)

Notes for end users

ALSA path configuration files can now be placed in user home directory

The code that loads the ALSA path configuration files now checks if the files exist in the directories specified with the XDG_DATA_HOME or XDG_DATA_DIRS environment variables (under pulseaudio/alsa-mixer/paths subdirectory). Those environment variables are defined by the XDG Base Directory Spec, and even if those environment variables aren't set, the XDG Base Directory Spec defines default locations that PulseAudio uses. In particular, in the usual case where XDG_DATA_HOME isn't set, the default value is $HOME/.local/share, so PulseAudio will look for path configuration files from that directory under the user's home. This is useful when it's necessary to customize the path configuration files. Previously the files in /usr/share/pulseaudio/alsa-mixer/paths had to be modified, and the modifications were lost whenever upgrading PulseAudio.

module-virtual-surround-sink rewritten

The convolution processing code has been rewritten to use a blocked FFT function, supporting potentially much larger impulse responses. Existing impulses will continue to function with the rewrite. The filter optionally supports asymmetrical responses, which may be enabled by specifying two sets of impulse files, using the hrir_left and hrir_right options instead of the original hrir option. Impulses should be matching format and channel layouts. The channel layout of the actual audio input for processing may be specified using the channel_map setting, so long as all specified channels are present in one or both impulses.

More options for module-jackdbus-detect

module-jackdbus-detect automatically loads a JACK sink and source when JACK starts. It's now possible to better configure the parameters of the sink and source. module-jackdbus-detect accepts these new arguments:

  • sink_name - Name for the JACK sink
  • source_name - Name for the JACK source
  • sink_properties - Additional properties to set on the JACK sink
  • sink_channel_map - Channel map for the JACK sink
  • source_channel_map - Channel map for the JACK source
  • source_properties - Additional properties to set on the JACK source
  • sink_client_name - The name that the JACK sink uses when it registers a client with JACK
  • source_client_name - The name that the JACK source uses when it registers a client with JACK

Improved support for HP Thunderbolt Dock 120W G2

HP Thunderbolt Dock 120W G2 has a 3.5 mm headset connector, and there's a separate audio module sold for the dock that provides speakerphone functionality (integrated microphone and speaker with echo cancellation). PulseAudio now has configuration that improves the device descriptions shown to the user and marks the speakerphone as a preferred device for voice audio.

Improved support for Behringer U-Phoria UMC22

Behringer U-Phoria UMC22 has two mono microphone inputs, which show up as a single stereo device in ALSA. PulseAudio now has custom configuration for this sound card, allowing the user to select a profile that enables only one or the other of the channels. That way recording applications don't end up recording a stereo track that has audio only on one channel.

Improved support for OnePlus Type-C Bullets

OnePlus Type-C Bullets is a simple USB-C headset device. It works with the default configuration too, but custom configuration was added to hide card profiles that don't make sense (for example, HDMI showed up as a supported profile with the default configuration).

Improved support for Sennheiser GSX 1000/1200 PRO

Sennheiser GSX 1000 and 1200 PRO are USB DACs designed for gaming. They have a separate mono output for voice audio, and they also provide a virtual surround device that downmixes 7.1 surround audio to headphones in hardware. Custom configuration was added to expose these devices properly.

New udev variable: PULSE_MODARGS

It's now possible to set module-alsa-card module arguments via udev configuration by using the PULSE_MODARGS variable. This allows for example setting tsched=false for a particular sound card that has trouble with the timer-based scheduling. Previously the options were to pass the tsched option to module-udev-detect in default.pa, causing all sound cards to be affected, or to give up automatic sound card detection and hotplug support altogether and configure each sound card manually in default.pa.

max_latency_msec argument added to module-null-source

It's now possible to configure the upper limit for latency of module-null-source with the max_latency_msec module argument.

module-filter-apply can take filter parameters from device properties

module-filter-apply automatically loads a filter module when a stream wants to be filtered (i.e. has the filter.want property). Previously additional filter module arguments could be provided by the stream in the filter.apply.<filter>.paramaters property, now it's possible to set filter.apply.<filter>.parameters also in the hardware sink/source properties, so a stream can simply request a filter without specifying the parameters, and the hardware device can set the (possibly hardware specific) parameters.

module-match can now be loaded multiple times

The original expectation was that there wouldn't be need for loading module-match multiple times, so it was marked as a module that isn't meant to (and can't) be loaded multiple times. However, it turned out to be useful to match against different properties with different configuration files, so the module can now be loaded multiple times.

Windows support added to Meson

PulseAudio's Meson build system was lagging behind the old Autotools build system when it came to building for Windows, now it should be possible to build PulseAudio for Windows with Meson using MinGW. There are also various fixes for Windows.

Notes for application developers

New API for sending messages from clients to modules

The new "message API" was added to make it easier to implement client interfaces in modules. This new infrastructure isn't yet used in any modules, so for now applications have no reason to start using the API. TODO: Better documentation is being worked on, link to the docs once they're available.

Notes for packagers

The startup script can now read additional configuration from the /etc/pulse/default.pa.d/ directory

PulseAudio's startup script (default.pa for the normal per-user mode and system.pa for the system mode) defines the set of modules to load automatically on startup. At the end of the scripts there is now an include directive that loads files from /etc/pulse/default.pa.d/ or /etc/pulse/system.pa.d/. This is primarily meant for distributions when they want to add some module to the configuration when installing a package: it's now easy to install a file containing the configuration for the installed package. Users can of course use this mechanism too if they prefer to add a module to the configuration without modifying the original startup script.

OSS support is now configurable in Meson

The oss-output Meson option can be used to explicitly enable or disable OSS support (previously, when using Meson instead of Autotools, OSS support was automatically enabled when the sys/soundcard.h header was found).

Valgrind support is now configurable in Meson

The valgrind Meson option can be used to explicitly enable or disable Valgrind support (previously Valgrind support was automatically enabled when the valgrind/memcheck.h header was found).

git shortlog

TODO