The annual Blizzard challenge has been launched. This isn’t some crazed track-meet for mad-scientist climatology. It is a competition – a sort of annual standardized test – for speech synthesis systems.
I could gabble on about speech synthesis and the exciting progress that is being made in the underlying technology, to help computers sound more natural and human-like. But I don’t want to bore you. (If you want a post on that, leave a comment. If I get any interest, I’ll post something.)
I think I’ll just mention that, aside from having a less grating voice in automated phone systems and whatnot, high-quality synthesis may be a real help for people with severe disabilities. You are likely familiar with Stephen Hawking, rock-star physicist and mathematician. You probably also know that he is wheelchair-bound and uses a computer to talk, due to a degenerative neural disorder.
He and many others, old and young, rely on computers to give their thoughts a voice. The better we get at producing human-like speech with computers, the more naturally these people will be able to interact with each other and with those of us who take fluent speech for granted.
So, I invite you to help make speech synthesis better. Your role in the Blizzard challenge is to rate the synthetic speech generated by the systems that have been entered in the contest. Just go here to start. (If you are a speech expert – such as a phonetician or speech technologist – then you need to use this link instead. It’s well-established that studying speech alters your perceptions in important ways. The task is the same, but the data will be analysed separately.)