Wikipedia is creating a crowdsourced speech engine

Wikipedia is developing a crowdsourced speech engine

Wikipedia introduced on Thursday that it’s collaborating with researchers from Sweden’s KTH Royal Institute of Know-how to develop an open, crowdsourced speech engine that may make the web encyclopedia extra accessible to individuals with studying or visible impairments. Wikipedia estimates that 25 % of its consumer base — roughly one hundred twenty five million individuals month-to-month — will profit from the brand new service. And whereas the engine can be optimized to be used on Wikipedia itself, any website operating MediaWiki software program will be capable of combine it as properly.

“Initially, our focus will probably be on the Swedish language, the place we’ll make use of our personal language assets,” KTH speech know-how professor Joakim Gustafson, stated in a press release. “Then we’ll do a primary English voice, which we anticipate to be fairly good, given the massive quantity of open supply linguistic assets. And eventually, we’ll do a rudimentary Arabic voice that shall be extra a proof of idea.” Wikipedia plans to have English, Swedish and Arabic prepared by the top of the yr earlier than shifting on to the opposite 280 languages its website helps.