DirectSound Driver Models

[The feature associated with this page, DirectSound, is a legacy feature. It has been superseded by WASAPI and Audio Graphs. Media Casting have been optimized for Windows 10 and Windows 11. Microsoft strongly recommends that new code use Media Casting instead of DirectSound, when possible. Microsoft suggests that existing code that uses the legacy APIs be rewritten to use the new APIs if possible.]

Under the VXD driver model, all DirectSound mixing is done in Dsound.vxd, a virtual device driver. Dsound.vxd also provides close access to the actual DMA buffer that the sound card uses to receive data from the host CPU. This is the same as the DirectSound primary buffer. A DirectSound application can set specific properties of the primary buffer, such as sampling rate and bit depth, thus changing the properties of the hardware itself.

Under the Windows Driver Model (WDM), DirectSound does not have direct access to the sound hardware, except in the case of hardware-accelerated buffers. Instead, DirectSound sends data to the kernel mixer. The kernel mixer's job is to convert the format of multiple audio streams to a common format, mix them together and send the result to the hardware. In a sense, it does what Dsound.vxd does. One major difference is that Dsound.vxd mixes only DirectSound buffer data, but the kernel mixer mixes all Windows audio data, including data from applications that use the Win32 waveOut functions. The rule that DirectSound and the waveform-audio output device cannot both be open at the same time is not true on systems with WDM drivers.

Of particular importance is the kernel mixer's relationship with the audio hardware. the kernel mixer is the only software on the system that can specify the format of the hardware's DMA buffer. It selects the format on the basis of sounds it is asked to mix. It sets the output format to the highest-quality format of sounds that it is asked to mix, or to the closest format to this that the hardware supports.

This has one very important implication: DirectSound cannot set the format of the hardware's DMA buffer. For your application, this means that the hardware format is based on the data you actually play. If you play a 44 kHz file, Kmixer will mix all data up to 44 kHz and ensure that the hardware is running at 44 kHz.

As an application developer, you don't choose the driver model used. That is completely determined by the type of sound card, the version of Windows, and the installed driver. For this reason, it is very important that you cover all possibilities when you test your application. DirectSound might be using Dsound.vxd or it might be using the kernel mixer, and you should ensure that your application's behavior and performance are acceptable on both.