[blueberry]

Is there a technical reason preventing the big guys (youtube/yahoo /vimeo etc.) from developing this kind of thing? I always thought running voice recognition on user submitted videos would require massive infrastructure. How are you guys planning to solve that? Congrats & best of luck with the rest!

[MediaSquirrel]

We're doing machine-augmented crowdsourcing, not straight voice recognition. It's different, and allows us to provide consistent high quality at low cost.

Essentially, we've built a crowdsourcing QA system a la Crowdflower, expect better and more suited to the specific task of video-to-text, and augmented it with speech recognition.

It's pretty fucking cool, if you ask me, but also very different from how people have traditionally approached this problem.

Also, the current site and product will be completely replaced in 8 days with something MUCH much better. So keep you eyes peeled.

-Matt