views:

444

answers:

3

I'm developing a web user interface to enter some information that is not very complex but needs to be loaded in real time. I think that the application could make use of speech recognition to facilitate the task.

Te core of the interface is being built with Javascript and jQuery, but can easily include a flash or silverlight component. I believe that´s probably the way to go...

I don't need to recognize everything that the user says, but only a few prerecorded commands. Also, I don't want the user to click on a button to specify the begining and the end of the spoken command. It should be detected live.

Is there anything that does this? I would be grateful if anyone tells me about a complete solution, free or commercial, as well as any advice on capturing a sound stream from the mic and process it with flash or sliverlight.

Sebastian.-

A: 

Check this

http://speechle.com/

Nickolay V. Shmyrev
It didn't work on any of my browsers... :(It says that I must allow access to the mic, but it doesn't ask anything. Anyway, I found the hidden flash component on the page and entered flash config window, so I already allowed the site to access to the mic, and reloaded the page, but the same message appears and nothing else happens.
Sebastián Grignoli
Now I see that the author said that this demo is down for maintenance. Sorry.
Sebastián Grignoli
+1  A: 

Hi

Nicolay told me about this. I made speechle as an example using the flash work I am doing. It is down for maintenance but other demos are working correctly. You can use the API i am working on at www.speechapi.com

Or try a demo at http://www.speechapi.com/live-demos/parrot.html

Some browser related issues I am trying to resolve but should work in firefox/chrome/safari. Seems IE isn't working at the moment but do try. Feel free to email me if you have issues.

ecret
Looks promising
Sebastián Grignoli
I've just tested the demo called "What", and I think this could really fit my needs.Could this technology be used to detect commands as they are spoken, without touching any start or end button (may be an end button only), and disregarding every unmatched sound?And last but not least, how are you licensing it?Great work, congratulations!
Sebastián Grignoli
Yes it has automatic mode where you don't need to click start and stop and it has speech detection where it ignores background sounds. Feel free to email me at the email address listed on the site.
ecret
Seems like Speechle works completely offline but Speechapi relays on a server for the recognition. Am I right?
Sebastián Grignoli
A: 

wami.csail.mit.edu

Ian