Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Dragon Naturally speaking way back had a way of training it to your voice and if mistakes were made you would type it in and from then on it would remember and recognize that item. Can't it be the same with these two-Evi recognizes regularly asked for call items and Siri goes into the web for more involved things-with trained recognition of voice and questions. Also how about having a real voice respond instead of computer voices to basic questions-there are only so many commonly used words so how hard can it be to decipher and put together those on the fly from recorded words with a real voice instead of this canned baloney?

No.

Dragon/Nuance only converts speech to text. It does not actually understand the meaning of said text. The training is for recognition of a word or some component of a word so that it can get the right text next time you said it.

From what I've read, Siri is a pipeline. The voice input is run though Nuance's speech recognition engine. So once Nuance converts what you say to text, then that text is fed into Siri's agent system for processing. The results are then printed on your iphone's screen and read back by a text to speech engine.

As for making a "real voice" by taking pieces of actual voice samples and putting it together? A lot harder than you think, as research in this has gone on for a very long time. You'll have problems with tones and inflection if you want realism and are using whole words. (amongst other things) Plus the sample database would be huge if you are using whole words, and limited to just those words.

By the way... http://www.telegraph.co.uk/technology/apple/8879705/The-voice-behind-Siri-breaks-his-silence.html
 
Last edited:
i just tried it and it suck.....boooo im so glad i have siri on my iphone 4 for free.. makes my wait fro iphone 5 much easier...
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.