Problem Solved:
FYI: it so happens that I had silence filler as
"sil" in my dictionary, but it was defined as "SIL" as an argument to bw algorithm. This was
causing bw to crash with a segmentation fault.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am trying to do my own training using SphinxTrain. So I follow the Manual, and try to do flat-start.
bw crashes with a segmentation fault after processing four files. I am trying to sanity
check to why it happens.
step (1) see is feature files are good.
step (2) see if gloal mean's/variance files are sensible.
According to the manual the variance files should be ASCII but they are not.... Does anyone know how to read them (like in Matlab).
How about reading the feature files?
-Try printp
-Can you send us a patch for the crash?
You can send it to me direct at archan at cs dot cmu dot edu.
Arthur
Problem Solved:
FYI: it so happens that I had silence filler as
"sil" in my dictionary, but it was defined as "SIL" as an argument to bw algorithm. This was
causing bw to crash with a segmentation fault.