[systemtrigger]

Another audio service I'd like to see: human powered text-to-speech...that scales.

Quick mockup: http://audiothink.com

Concept: User provides prose and a narrator reads it. User pays based on the number of words. Narrator posts audio files to user's private podcast. User subscribes to podcast or dials a (Twilio-powered) phone line. Service uses Mechanical Turk workers to do the narrating. But that introduces hard quality control issues. How does one automate the outsourcing of...art? I thought maybe if I set up a good qualification and reward system the service could have quality safeguards - but that's hardly a solid foundation on which to devote months of hacking. So that's where I stopped.