I've successfully used the -kws in pocketsphinx with good results but I need to train my own model for 5 words using shinxtrain.
My database is not huge and is manually collected.
First issue I ran into was the "not enough data" so I duplicated my database to reach the required 1 hour.
But still I got bad accuracy:
TOTAL Words: 1260 Correct: 396 Errors: 864
TOTAL Percent correct = 31.43% Error = 68.57% Accuracy = 31.43%
Is there a special way to train small models for few words?
PS: I cannot use model adaptation.
Thanks.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hey,
I've successfully used the -kws in pocketsphinx with good results but I need to train my own model for 5 words using shinxtrain.
My database is not huge and is manually collected.
First issue I ran into was the "not enough data" so I duplicated my database to reach the required 1 hour.
But still I got bad accuracy:
TOTAL Words: 1260 Correct: 396 Errors: 864
TOTAL Percent correct = 31.43% Error = 68.57% Accuracy = 31.43%
Is there a special way to train small models for few words?
PS: I cannot use model adaptation.
Thanks.
it doesn't work like that.