I am using pocketsphinx 0.6 and all the words are in the dictionary.
Btw the above problem is solved by adding
FSG_BEGIN second
to the fsg file as per the example in the source code tidigits.fsg.
Now I have a new problem the accuracy of the words spoken while the second fsg
is loaded is low, the accuracy is low when he new fsg has more states than the
first.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I am having problem changing the FSG file dynamically in pocketsphinx using
gstreamer. I am able to initialize the fsg for the first time.
When I attempt to change the FSG using
it gives
INFO: fsg_model.c(542): FSG: 50 states, 44 unique words, 44 transitions (89
null)
INFO: fsg_model.c(177): Computing transitive closure for null transitions
INFO: fsg_model.c(228): 43 null transitions added
INFO: fsg_model.c(281): Adding silence transitions for <sil> to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++NOISE++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++BREATH++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++SMACK++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++COUGH++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++LAUGH++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++TONE++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++UH++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(281): Adding silence transitions for ++UM++ to FSG
INFO: fsg_model.c(301): Added 50 silence word transitions
INFO: fsg_model.c(325): Adding alternate word transitions (ZERO,ZERO(2)) to
FSG
INFO: fsg_model.c(358): Added 1 alternate word transitions
INFO: fsg_model.c(325): Adding alternate word transitions (ONE,ONE(2)) to FSG
INFO: fsg_model.c(358): Added 1 alternate word transitions
INFO: fsg_model.c(325): Adding alternate word transitions (A,A(2)) to FSG
INFO: fsg_model.c(358): Added 1 alternate word transitions
INFO: fsg_model.c(325): Adding alternate word transitions (EXIT,EXIT(2)) to
FSG
INFO: fsg_model.c(358): Added 1 alternate word transitions </sil>
and then simply exits.
The source code of pocket sphinx gstreamer plugin does allow this change. I am
not able to get the mistake.
Please help.
I am using pocketsphinx 0.6 and all the words are in the dictionary.
Btw the above problem is solved by adding
FSG_BEGIN second
to the fsg file as per the example in the source code tidigits.fsg.
Now I have a new problem the accuracy of the words spoken while the second fsg
is loaded is low, the accuracy is low when he new fsg has more states than the
first.
Thanks every one for the help. I solved my problem, the problem was that I was
not running
after the g_object_set.
That issue was fixed after the release, will be available in next version
Hm, that's not a straightforward API indeed. We need to fix this.