Melgrave (Australia) – A beta tester working with build 5472, the July Community Technology Preview of Microsoft Windows Vista, has posted an MP4 video of his own desktop, showing an apparently successful demonstration of Vista’s new speech recognition features. The new alternative input model, which may become standard in the final editions, allows voice input to be used in place of mouse motions, so that each graphical device has a voice-enabled shortcut.
In the video, you can clearly see how Vista responds in situations where it doesn’t quite understand the command, or when it cannot resolve the term to which the speaker is referring. In such instances, Vista highlights all the possible controls to which a command may have referred, with blue, semi-transparent overlays, each with numbers. You read the number to tell Vista which control you meant to activate. Vista can apparently also analyze textual contents of Web pages, so that the user can speak the contents of links – such as “an apparently successful demonstration” above – to activate them.
When a spoken command doesn’t seem to refer to a button or active control, Vista assumes the user wants to type something. So if a program such as Notepad or Word is open, it begins responding to dictation. The demonstration also shows how Vista will handle the sticky topic of voice-directed cursor movement for painting and drawing programs. Whether you want to describe this part as “successful” depends on your point of view.