I am sorry Nickolay. Most of the times it predicts more phonemes than what's in the original word. And in fact most of them are predicted incorrectly. The predicted time stamps are somewhat okay to some extent compared to the phonemes tho.
Last edit: Vikram Natraj 2019-03-29
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
https://github.com/cmusphinx/pocketsphinx/blob/master/swig/python/test/phoneme_test.py
Why does the above code give terrible results??
Should I use another model to train it instead?
Please provide the link for the latest model on sourceforge.net if that is available.
Thank you
What exactly is "terrible results" for you?
accuracy: actual phonemes vs predicted phonemes
It doesn't match and i mean to a great extent.
It matches for me here
I am sorry Nickolay. Most of the times it predicts more phonemes than what's in the original word. And in fact most of them are predicted incorrectly. The predicted time stamps are somewhat okay to some extent compared to the phonemes tho.
Last edit: Vikram Natraj 2019-03-29