Yeah, I well know that Microsoft has made it difficult. In fact that's exactly why I'm trying to interest someone in making it easy. :-) The thing that makes it stand outside the heuristics of that philosophy is that it is not for high-performance applications, but just for giving users a better listening experience. The path I use now has more latency than I think would be incurred by this application. Currently I use Virtual Audio Cable feeding the VSTHost Windows application which is hosting a modular audio plugin, PlogueBiduleVST, do do the heavy lifting and it doesn't put audio out of sync with video to a significant degree. What I'm proposing would be similar but without the Virtual Audio Cable latency. While highly desirable I don't know yet whether or how some of the controls for an inner modular filter implementation could be mapped out to the user and that could be a significant problem. The actual tweaking can be done across a process boundry with the OSC protocol. I've tested that with PlogueBiduleVST used as the inner modular host and a smart phone OSC controller but presenting a control surface within Windows that uses OSC could be hairy and I don't really want to limit the choice of an inner processing plugin to PlogueBidule due to its proprietary nature. On Wed, May 8, 2013 at 10:57 AM, Tim Roberts <timr@xxxxxxxxx> wrote: > Don Gateley wrote: > > > > My reason for wanting a driver/device interface is not to do any low > > level kernel processing but simply to give a user a familiar way to > > direct output to a DSP filter via the "Windows Sound" control panel > > and to configure the plugin and real output device via it. Is there a > > path to such a driver interface that doesn't involve kernel mode > > (other than perhaps as a hidden bridge)? I clearly misdirected my > > query. Can you suggest a forum or list more appropriate? > > I don't think you misdirected your query at all. The people who would > know how to do this live on this list. > > One issue is that you're up against a philosophy problem. The > overarching mantra in the redesign of the Audio Engine in Vista was > efficiency and low latency. On a system with an HDAudio device (which > is most systems these days), audio traffic never passes through kernel > mode at all. The Audio Engine process writes directly to the board's > audio buffers. > > Microsoft has made it difficult to insert arbitrary helpful "filters" in > the audio path, because that makes the path non-deterministic for > high-performance audio applications. What you are doing here is counter > to that trend. Instead of app to audio engine to device, you'll now > have app to kernel interceptor to user-mode VST host to VST graph to > audio engine to device. That's a lot of players. > > I'm not saying this is impossible, just that this is a somewhat bigger > problem than you may realize. I see your concept, and I think it could > be made to work, but it's a big job. > > -- > Tim Roberts, timr@xxxxxxxxx > Providenza & Boekelheide, Inc. > > ****************** > > WDMAUDIODEV addresses: > Post message: mailto:wdmaudiodev@xxxxxxxxxxxxx > Subscribe: mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=subscribe > Unsubscribe: mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=unsubscribe > Moderator: mailto:wdmaudiodev-moderators@xxxxxxxxxxxxx > > URL to WDMAUDIODEV page: > http://www.wdmaudiodev.com/ > >