As explained earlier, the trouble is that the framework Scrivener uses isn’t supported by DNS, so there needs to be compatibility there to enable full interoperability between the programs. We do hope to be able to offer some sort of voice recognition support in the next major upgrade, whether with DNS or the built-in Windows software–both, if possible, certainly–but working around the current limitations is not an easy task. The next major release will use an updated version of the framework which may offer some new options (still not full support, sadly) and we’ll be taking the time to try to hammer out compatibility if at all possible.