I have just created on ASR in english language with my voice only. I have used 9 hrs of training data. and used 1 hr of test data. The accuracy is coming 91.03%.
When I used the system giving something around 8 minutes of my voice file, the decoding was not that good. [I have segmented the 8 minutes file using sox and then decoded using sphinx3 decoder.]
Some segments are properly recognized but some are meaningless.
Is there anything I missing with.
Please comment.
Thanks much.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I have just created on ASR in english language with my voice only. I have used 9 hrs of training data. and used 1 hr of test data. The accuracy is coming 91.03%.
When I used the system giving something around 8 minutes of my voice file, the decoding was not that good.
[I have segmented the 8 minutes file using sox and then decoded using sphinx3 decoder.]
Some segments are properly recognized but some are meaningless.
Is there anything I missing with.
Please comment.
Thanks much.
You need to provide the data file and other information in order to get help on the accuracy issues.