cat 2929.dic
BACKWARD B AE K W ER D
FORWARD F AO R W ER D
LEFT L EH F T
RIGHT R AY T
STOP S T AA P
TURN T ER N
If I say a word included the 2929.dic,it worked fine. the accuracy is above 90%.
But if I say a word excluded the 2929.dic the accuracy is so poor.
e.g, I said "acquire".Recognition results: "BACKWORD". I said "world" Recognition results: "FORWARD".
How can i solve this problem?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you for your kind help.
I don't want a big language model and dictionary. I just need a samll them which
include several command words. I have a ipcamera. I want if I say a command to it,such as turn left, in respon,it can do a action. if I say a wrong command to the ipcamera it will reject to aciton.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm runnig demo app of pocketsphinx:
pocketsphinx_continuous -inmic yes -lm 2929.lm -dict 2929.dic
cat 2929.dic
BACKWARD B AE K W ER D
FORWARD F AO R W ER D
LEFT L EH F T
RIGHT R AY T
STOP S T AA P
TURN T ER N
If I say a word included the 2929.dic,it worked fine. the accuracy is above 90%.
But if I say a word excluded the 2929.dic the accuracy is so poor.
e.g, I said "acquire".Recognition results: "BACKWORD". I said "world" Recognition results: "FORWARD".
How can i solve this problem?
Include all words you need in the language model and in the dictionary
Thank you for your kind help.
I don't want a big language model and dictionary. I just need a samll them which
include several command words. I have a ipcamera. I want if I say a command to it,such as turn left, in respon,it can do a action. if I say a wrong command to the ipcamera it will reject to aciton.
You can use keyword spotting for continuous listening for activation keyword. See for details
http://cmusphinx.sourceforge.net/wiki/tutoriallm