Audio on Linux explained: ALSA, Pulse, Jack, Pipewire…

28 Dec 2021

This article is a small, simplified explanation of audio on Linux. Let's understand audio on Linux!

• The Soundcard: in your computer, the soundcard is responsible of converting digital information, understood by the operating system, to analogic signal, used by devices (a jack headphone or microphone), and vice-versa.
• The operating system, by itself, doesn't know how to read/write bytes to the soundcard. It needs a driver.
  • In the early days of audio on Linux, the driver was OSS (Open Sound System), but it had some limitations and became proprietary.
  • Now, the commonly used driver is ALSA (Advanced Linux Sound Architecture).
• With these drivers, it is possible for an application to use the soundcard: VLC can output directly to ALSA.
• However, these drivers have an important problem: they don't handle multiplexing. This means only one application can use the soundcard (microphone or speakers) at a time.
• This problem is solved by using a sound server: it merges all audio streams into one, which it then sends to ALSA.
• The sound server is also responsible of resampling the audio. If two streams use a different frenquency, let's say 44.1kHz and 48kHz, the server will “convert” (resample) these streams to the frequency of it's configuration, 44.1kHz for example.
• The commonly used sound server nowadays is PulseAudio, which provides an abstraction layer to the applications. From the application point of view, they're just reading/writing to PulseAudio.
• PulseAudio even allows using multiple sound cards at once: the integraded soundcard in your computer, a USB microphone, the HDMI audio output of your graphics card, … This can be easily configured in pavucontrol. See picture below.
• PulseAudio has two main limitations.
  • It hardly handles managing audio as a graph (eg. having an applications streaming to two sound cards).
  • It is not real-time, there's a small latency of 40-80ms.
  • These two drawbacks are irrelevant to the majority of users, but…
• JACK, an alternative sound server, solves these problems. It's designed for the audio profesionnals: very low latency, flexible audio connections (with graphs if you install qjackctl), even MIDI support for instruments.
  • There are two “versions” of JACK: JACK and JACK2. JACK2 is written in C++, whereas JACK is written in C.
  • Most applications are written for PulseAudio. If you run JACK for a specific app (Ardour, for example), you will notice the other apps trying to emmit sound are like frozen. Indeed, the soundcard is used by JACK, which means PulseAudio cannot read/write to it. This problem can be solved by adding these commands to the adequate events in qjackctl. See picture below.
• That's quite a lot of programs for audio on Linux… Now, let's introduce Pipewire. Pipewire combines the simplicity of PulseAudio and the advantages of JACK.
  • Pipewire can act as a drop-in replacement for PulseAudio by installing the pipewire-pulse addon. This should be transparent to apps, as they “think” they are interacting with PulseAudio.
  • The same applies to JACK by installing pipewire-jack. However, you might need to run JACK apps with the pw_jack command: eg. pw_jack ardour6.
• We've talked about audio… video on Linux is still at the “driver step”: a webcam on Linux can only be used by one app at a time. There is no “video server” which handles multiplexing video… Well… there wasn't. Pipewire also introduces video multiplexing and native screen-sharing under the wayland display server.

And… That's it! Hope it helped you get a clearer understanding of audio on Linux.

pavucontrol

qjackctl compatiblity