Compressed data support in sound subsystem

This proposal has been rejected.


One Line Summary

The new streaming interface proposal provides a framework in sound subsystem to support offloaded decoders and encoders


To reduce power consumption, the majority of recent SoCs rely on audio DSPs or dedicated low-power cores. Audio codec vendors have likewise started to embed digital audio processing capabilities in their mixed-signal chips. Support for such hardware is however limited in Linux due to strong assumptions in the ALSA API; compressed data is currently limited to IEC61937-formatted bitstreams such as Dolby Digital and DTS. Native support for popular compressed formats isn’t possible with the PCM interface only manipulate samples instead of bytes.

In this presentation, we provide an overview of a new streaming interface proposal to support offloaded decoders and encoders. To avoid re-inventing the wheel and enable easy modification of existing middleware, the parameters listed in the API are compatible with existing standards such as OpenMAX IL. We detail the required links with ASoC and DAPM for power management. UCM extensions to expose devices with compressed data support to applications are also discussed


audio, alsa, sound, ASoC, compressed path, offload


  • Pierre Bossart

    Intel Corporation


    Audio and system architecture related positions since 1997 with Digigram, STMicroelectronics, Freescale and Intel.

  • Vinod Koul

    Intel Corp


    Vinod Koul is a kernel developer mostly focused on embedded audio and slave-dma engine.
    He currently is the maintainer for the slave-dma engine.