Voce Library / Speech Recognition
in
Contributed Library Questions
•
2 years ago
Hi Guys,
following this thread from the old forum
I started to play around with Voce
and it seems to work even though not very accurate.
Now I would like to understand a bit more what's actually happening behind and how I can implement own grammar files (.gram) which tetermin what the application can understand etc.
Therefor I would like to start discussing here and would be very happy if people join in who already used / experimented with this lib.
My first concern would be how to design the grammar file (which software, which structure is best). I am adding here one file I have been using from the examples of Voce. It works quiet ok with "quit" and "reset" but it always also gets some "ghost input" which is a mix of other words.
- #JSGF V1.0;
- /**
- * JSGF Digits Grammar file
- */
- grammar objects;
- public <objects> = <colors> <objectTypes>;
- public <miscellaneous> = (quit | reset);
- <colors> = (yellow | red | blue | green | purple | orange);
- <objectTypes> = (box | sphere | wall | tower | character);
3