Sound API Ubuntu Linux

Question

I want to write some native C/C++ code that takes some uncompressed PCM audio data and plays it out through my speakers on Ubuntu Linux 12.04. Can you tell me what are the different default audio APIs/libraries that are installed? What is their low-level architecture and how do they interarct with the kernel? (which syscalls)

(I've heard different terms like ALSA, PulseAudio, OSS, and so on, but I don't really have a clear picture of which does what and how they fit together.)

Answer 1

On Linux, you have LOTS of Linux audio APIs to choose from. Here is a graph I made a few years ago demonstrating the relationships among the various Linux audio APIs.

However, if you are only targeting Ubuntu Linux and using C/C++, and you don't need anything too fancy (just taking raw PCM data and shoving it out to the speakers), I would recommend using PulseAudio and sticking to the "simple" API. A trivial example can be found here. It's the path of least resistance to getting a proof of concept working.

Answer 2

In very simple terms, ALSA provides the low level infrastructure for audio in Linux, and Pulse provides the higher (more desktop-friendly) level. As far as most mainstream Linux distributions are concerned (Ubuntu included), OSS is obsolete.

You ask about syscalls. You don't use audio in Linux via syscalls. If you choose to use ALSA directly (not a good idea for simple audio playback), you'd employ its userspace library libasound2 (even OSS would be used via /dev/ files rather than syscalls).

If you are targeting modern Linux distros only, consider using the Pulse API. If you want more flexibility as to what distro you are targeting and the potential for cross-platform support, you could try a library which abstracts away the OS-specific audio API - for example, PortAudio (http://www.portaudio.com/).

If you've never done any audio development, you may find a library such as PortAudio easier to work with than addressing PulseAudio directly.