Its been a bit quiet on this topic recently but I have been looking into how Windows Core Audio works at a low level so I thought I would share what I think I now know ?
Audio apps normally work with calls to a higher level API but one way or another they must open an audio stream session. The session requires information on the number of streams, their bit rate, the location of the audio apps output buffer, etc. The audio session manager will setup an audio end point buffer in system memory and inform the audio engine of its location. The audio engine can read data from many end point buffers and mix it together with Windows efects before outputting the result to an audio devices hardware buffer via its driver.
An audio app can also request exclusive use of an audio device and if granted the session manager will send stream packet data directly to the audio device, bypassing the audio engine.
Exclusive use only allows the playback of one stream at a time but it can contain multiple channels. Exclusive streams also get marked as pro audio so they recieve a priority boost too. No other sounds, including system sounds, can be played on a device that is in exclusive mode.
.
