Proposal for MetaLanguage Development

ejTalk believes that there are many benefits with an open-like source approach. Many contributors with various points of view leveraging off the synergy of the combined effort. But most true open source projects that reach "critical mass" and lead to a real work product are mass market concepts. They range from the obvious (Linux) to the cultural (GAIM - generic IM, mp3 rippers) to more specific power tools (video and audio editors). The speech development domain doesn't fit with these projects because only a few of us in the industry can understand what we are missing and why it is important. How do we get the ball rolling?

ejTalk will organize a project that will specify, prototype and document a working MetaLanguage for speech applications. This effort will be focused on working prototypes and related tools that demonstrate the functionality of this new metalanguage. It will evolve directly from the needs and examples of the MetaLanguage membership. It will progress a rate commensurate with the membership funding.

One of the goals of this seed activity could (and probably should) be to convert this project into a full-fledged open source project. A period of one to three years should be enough time to demonstrate functionality and foreshadow the value of the methodology. This in turn will attract true open source contributors from a much larger pool of speech application developers.

One of the core tenants of this project must be to maximize the amount of useful technology produced given the limited resources. The best way to do that is to develop many small demonstrable components across the range of the domain of a MetaLanguage rather than picking one course to which all members must "sign on". This is not an "end all" project but more of an organic exploration of many of the most readily tackled facets of the MetaLanguage with working, albeit primordial, components. The components that stir the most interest among the members will create a pragmatic selection process for further work. This basic yet functional "pool of components" will also inspire new and fresh ways to look at the task of building a speech application.

Interesting. How might it work?

Here are some ideas