Which do people find best, using a Language Model or a Grammar file? If you specify
both does pocketshpinx use both?
I have both but I find that the grammer file is substituting extra words in between/after
recognised words that does not happen while using the language model.
example. 'display default' is correctly recognised with the language model
'display default UP' with the grammar file.
Overall, modern speech recognition interfaces tend to be more natural and avoid command-and-control style of previous generation. For that reason most interface designers prefer natural language recognition with statistical language model than old-fashioned VXML grammars.
Tutorial also lists other considerations to take into account.
If you specify both does pocketshpinx use both?
It will use language model which has more priority
I have both but I find that the grammer file is substituting extra words in between/after
recognised words that does not happen while using the language model.
This is because your grammar ensures last UP, it doesn't allow skip. You should have used * instead of + in (up|down)+ to allow to skip last word.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I want to use RNNLM with sphinx..
Can i get to know how can rnnln language model be adapted to sphinx environment. Any suggestion or help related to it would be very useful..
Thanks in advance for replying
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi all,
Which do people find best, using a Language Model or a Grammar file? If you specify
both does pocketshpinx use both?
I have both but I find that the grammer file is substituting extra words in between/after
recognised words that does not happen while using the language model.
example. 'display default' is correctly recognised with the language model
'display default UP' with the grammar file.
grammar: <display_cmd> = display (default | up | down) (up | down)+;</display_cmd>
as 'display' can be: display default, display up, display down, display up/down recurring....
Thanks
Russ
From http://cmusphinx.sourceforge.net/wiki/tutoriallm
Tutorial also lists other considerations to take into account.
It will use language model which has more priority
This is because your grammar ensures last UP, it doesn't allow skip. You should have used * instead of + in (up|down)+ to allow to skip last word.
Once again, a big thank you for the info.
Much appreciated Nickolay
Russ
Hi..
I want to use RNNLM with sphinx..
Can i get to know how can rnnln language model be adapted to sphinx environment. Any suggestion or help related to it would be very useful..
Thanks in advance for replying
Not sure why you are writing this as a comment and not as a new qustion...
RNNLM can be used for re-scoring N-best list or lattices. Probably some file format conversion will be needed