Virtual Worlds and Speech Recognition

An article emerged recently in Speech Technology Magazine concerning a topic of usual interest in this space — the use of virtual worlds among people with disabilities. Among the examples cited are cases where participation improves hand/eye coordination, socialization skills and cognitive functionality.

An interesting angle being additionally explored is the challenge for users who rely on speech recognition or dictation software to access their computers. Second Life’s chat capabilities are accessible via Nuance’s Dragon NaturallySpeaking (formerly called Dragon Dictate). However, not all features are supported and Windows incompatibilities exist:

Because Second Life does not use standard Windows application programming interfaces, not everything works by voice. For example, Dragon’s Select-and-Say feature is not available in the chat window, and menu items do not have assigned shortcut keys, so users cannot write keyboard shortcuts to voice-enable the environment. Dictators can write scripts to control the mouse or memorize MouseGrid coordinates to control the features they use most often, but when the graphic user interface changes, the customization becomes worthless.

The good news is that Linden Labs is continuing to develop prototypes that take into account users with differing abilities. As mentioned before in this space, Linden is exploring access through various means of input and the feedback mechanisms that guide a user’s experience.

An item that is clearly of interest and need would be a standard taxonomy applied to environments and objects. Public accommodation policies germinate from accepted standards that are applied to multiple contexts. A possible solution might be introduced through the Dublin Core Metadata Initiative, an organization dedicated to developing baseline metadata practices to a wide range of applications.

As mentioned by open source and accessibility advocate Matthew Smith, Dublin Core could be extensible to other taxonomies by pointing descriptions and designations to objects within a virtual environment. The question is how and where that technical handshake would take place, and whether that interoperability could be extended to multiple platforms.

Also of interest would be whether metadata standards could be transitioned from the 2D to the 3D web spaces. This would allow items within virtual worlds to be more readable not only to assistive devices, but also text-based media such as search engines and RSS feeds. The opinion here is that such developments need to be explored if virtual worlds are to attain further transactional, commercial and social value to nondisabled populations.