Hi ! just an idea i was thinking to give robot the ability to control 100% with voiciereconition
webkitspeech reconition is just perfect but can be ajusted
What do you think about something like this : https://www.sparkfun.com/products/13316 ( maybe exist something more autonomous )
The idea / problem :
When no sound is detected webkitspeech reconition is off and we need to click on the microphone.
In noisy environement webkitspeech is very long to send command
So , I tell this hardware voice comand ( or an other thing ):
"listen" > the command ear.startListening() is send
"Stop listening" > In a noisy environement the webkitspeech reconition ear.stoptListening() ( or ear.send data i don't know if exist )
Borg it in!
Speech recognition is difficult. webkit is great at general speech recognition across many languages, but it doesn't work unless you have an internet connection. Sphinx works is you don't have an internet connection, but it has a very limited vocabulary for what it will recognize.
I think having more options for speech recognition in MRL would be a good thing. It's pretty easy to create a new service in Java that implements the SpeechRecognition interface, and you should be all set to use it.
It looks like this is a device that connects to the arduino, so after the next round of refactoring of MRLComm it will be easier to implement. It looks like it supports maybe as many as 300 commands. So it's not general speech recognition, it's more like sphinx that has a fixed grammer when it starts listening. I would say, if you're interested in using it, get one test it out, let us know and we'll help you borg it in!