hi
i have model trained "sphinxtrain" using a corpus of 85 hours of french broadcast news , for decodeing i use a corpus of 1 hour i have WER = 29 % ,
i wanna do an mllr adaptation for telephonic speech
to start with one itération of baumwelch i had errors :
ERROR: "backward.c", line 409: final state not reached
ERROR: "baum_welch.c", line 300: 19981207_0700_0800/19981207_0700_0800.241 ignored
this error ocuur for all utterances 90% of the telephonic corpus ,
can some body has any idea ?
thank
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
hi
i have model trained "sphinxtrain" using a corpus of 85 hours of french broadcast news , for decodeing i use a corpus of 1 hour i have WER = 29 % ,
i wanna do an mllr adaptation for telephonic speech
to start with one itération of baumwelch i had errors :
ERROR: "backward.c", line 409: final state not reached
ERROR: "baum_welch.c", line 300: 19981207_0700_0800/19981207_0700_0800.241 ignored
this error ocuur for all utterances 90% of the telephonic corpus ,
can some body has any idea ?
thank