I need to code a 'dispatch driver' that actually allows a standard
microphone driver to continue performing its duty (reading the
microphone) but that will be able to have multiple sessions connected
to it at the same time.

As an example, application A opens a session with a microphone,
application B also opens a session with the same microphone. While the
user speaks without pressing any key, audio goes into application A.
But, when the user holds the ALT key down, audio is sent to
application B (while application A gets silence data).

I thought of coding this by producing a 'dispatch driver' between the
application and the 'physical driver' that talks to the physical
device itself, which would allow multiple connects while scaning the
keyboard and directing the audio data to only one application
depending on the keyboard status.

Does anyone have any sample code that could help me in that area
(Driver X calling Driver Y)? Any reaction to that approach? Does that
appear to be feasible, or am I dreaming?

Thanks.

More questions:

Is it possible to create in intermediate driver over sound input
drivers? Then use external binding in order to connect to the driver
that communicates with the physical device? Or, are intermediate
drivers reserved for network devices and limited to them by design?
What about NDIS and SAP, should I look at that?

Thank you.