when I train about an hour of data using SphinxTrain for 1217 word in the dictionary,I obtained 15.78 over all liklihood. is it ok?
When I run the batch test the error rate was 20 %
if I create a jar out of this training result can I use it in real application?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you Arthur, when I increase the training data likelyhood goes down is ok
But if I increase training data but with the same set of word but different speaker the likelihood increases is it ok
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Your training is ok and the error rate makes sense. You could probably create a simple application out of it. If you want to build something which has good performance, you do want to have more training data.
Arthur
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
when I train about an hour of data using SphinxTrain for 1217 word in the dictionary,I obtained 15.78 over all liklihood. is it ok?
When I run the batch test the error rate was 20 %
if I create a jar out of this training result can I use it in real application?
Thank you Arthur, when I increase the training data likelyhood goes down is ok
But if I increase training data but with the same set of word but different speaker the likelihood increases is it ok
Your training is ok and the error rate makes sense. You could probably create a simple application out of it. If you want to build something which has good performance, you do want to have more training data.
Arthur