Driver Atlas

Audio-driver lifecycle reference

The layers that carry audio from an application-facing API to playback, capture, and control endpoints.

Audio-driver lifecycle reference

Audio is a chain of representations

An application works with a stream described by samples, channels, timing, and format. Audio services and APIs represent that stream for the host audio stack, while a device-facing component translates it into the transport and hardware contract. “Audio driver” can therefore name one layer in a larger path.

Playback and capture are different directions

A physical audio device may expose playback endpoints, capture endpoints, or both. The endpoint is a logical source or destination and can have its own format and state. A single product can present several interfaces without each interface being a separate physical device.

Format and topology have scope

Sample rate, channel layout, bit depth, and encoding describe a stream at a particular boundary. Device topology describes connections among audio nodes, controls, and endpoints. Neither a stream format nor a topology record should be treated as a universal description of the hardware.

Lifecycle records are layered

Enumeration, package matching, service activation, endpoint publication, stream negotiation, and runtime delivery are distinct lifecycle moments. A version record may identify a package or component while an endpoint state describes what the audio subsystem currently exposes.

Reference facts

Core layers
Application API, audio service, driver, transport, and device endpoint
Directions
Playback, capture, and control can be separate interfaces
Stream fields
Samples, channels, timing, rate, depth, and encoding

Questions and answers

Is an audio endpoint the same as a speaker?

No. An endpoint is a logical playback, capture, or control destination or source. It may be backed by speakers, a headset, a microphone, or another device.

Does the audio driver choose every stream format?

Format negotiation is shared across the application, audio API, services, driver, and endpoint capabilities. No single label necessarily owns every choice.

Why can one audio product have several entries?

A product may expose playback, capture, control, transport, and vendor interfaces with distinct software roles.

Further reading