| Thread Previous • Date Previous • Date Next • Thread Next |
On 13/02/12 12:59, supernova wrote:
yeah, it could do, fairly easily I think, the menu fragments need to be reformed into a jsgf grammar and then passed to pocketsphinx to listen for. Continuous speech recognition is still some way off, but if there is a grammar that it is listening for (i.e. it matches against a limited vocabulary rather than the entire English language) then it can be really quite good. The menu data that the HUD accesses appears to be quite a good corpus of stuff to feed into a vocabulary file. I think now I know a bit about dbus introspection that the menu data is in there somewhere, I am fairly sure all the bits exist today and just need plumbing together.Let's focus on HUD innovation, leaving the menu as a backstop till we have great solutions for *both* aspects of the menu; exploration/discovery, and command invocation.this is important. HUD has potential, just ad the dash. They are the real tutor of any Ubuntu user. Maybe in the future HUD will understand sound commands?! Supernova
Alan.
| Thread Previous • Date Previous • Date Next • Thread Next |
