I'm trying to develop an application that recognizes Arabic numbers using Sphinx-4. I used SphinxTrain to create my model. I then Modfied the demo application (hellodigits) to test my model. My modification included changing the front end to the mfcfrontend to fit with the features generated by SphinxTrain. When I run the demo, the recognizer seems to work forever (it does not even return null). And when I change the logging level to the "FINSET", I get all the searchmanager results as the following, with this strange -? value.
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,TH]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,TH]>_P(eight[V,EY])-G190
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,F]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,F]>_P(eight[V,EY])-G190
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,N]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,N]>_P(eight[V,EY])-G190
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,HH]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,HH]>_P(eight[V,EY])-G190
What could be the reason for this?
And also, what other training utilites could be used to generate models for Sphinx-4?
Regards,
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello there,
I'm trying to develop an application that recognizes Arabic numbers using Sphinx-4. I used SphinxTrain to create my model. I then Modfied the demo application (hellodigits) to test my model. My modification included changing the front end to the mfcfrontend to fit with the features generated by SphinxTrain. When I run the demo, the recognizer seems to work forever (it does not even return null). And when I change the logging level to the "FINSET", I get all the searchmanager results as the following, with this strange -? value.
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,TH]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,TH]>_P(eight[V,EY])-G190
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,F]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,F]>_P(eight[V,EY])-G190
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,N]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,N]>_P(eight[V,EY])-G190
10:32.133 FINER searchManager BT 0028 -? 0.0000000E00 0.0000000E00 0.0000000E00 S2_U1<T[EY,HH]>_P(eight[V,EY])-G190 for state S2_U1<T[EY,HH]>_P(eight[V,EY])-G190
What could be the reason for this?
And also, what other training utilites could be used to generate models for Sphinx-4?
Regards,
"I get all the searchmanager results as the following"
I mean, similar to the following :)