SOUND plans to move more audio functions in the kernel
by admin ·
A previously unknown author has announced a new audio system that supports all major audio functions in the Linux and FreeBSD kernels and a simple shift, wants to offer OSS-oriented interface for applications.
SOUND – kernel level audio Next Generation says on a website that has published an unknown author. The author is dissatisfied with the separation between kernel drivers (ALSA or OSS4) and audio routers like PulseAudio or JACK. He contrasts the new audio system, low latency, full latency compensation and provides the routing of audio streams between endpoints, and there are at the end points of view of applications do not distinguish between hardware and processes.
The reason for the design of a new audio system for the author that the existing systems have problems with the latency and the synchronization of audio and video. The audio processing belong to the kernel, because even latency of 4 ms, the sound effects heard. After all, he admits that the current separation between kernel and user processes in spite of the necessary context switching works quite well, at least as long as only a few processes involved. For professional applications, however, he sees the limits are reached.
SOUND will also increase the level of abstraction for applications. The author believes that the existing systems such provide only the ALSA hardware interface for the applications. This is comparable to that network would have to construct applications, the TCP packets themselves, to be sent over the network.
The author sees the audio mainly as a routing problem where the sources and sinks must be connected. Audio applications will work with audio, simply by opening a file descriptor, this with ioctl configure and read and write to make inputs and outputs. This is essentially the API OSS4 (now the free Open Sound System). But according to the author OSS4 provides no power management, making it unusable for mobile devices, and no MIDI. SOUND is to contain it all.
SOUND is according to the author in development, but it is still too unstable and experimentally for the first publication. The publication is intended as a free software once it is stable in the broad and a native driver for HD audio chips from Intel is ready. The author would like to introduce the system later in the kernel, but would need new drivers for all audio chips can be written. Perhaps it would not even have much work to deduce this from the existing driver for ALSA or OSS, but for predictions, it is too early.