My transcriptions are manually developed and they contains the correct
pronunciation variants from the dictionary. During MMIE training in module
number 60.lattice generation, I found out that the force alignment estimate
different pronunciation from the dictionary and they are not accurate 100%. I
want to use my transcriptions and need the force aligner to estimate only the
timing in the wdseg files and keep my original variants.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
My transcriptions are manually developed and they contains the correct
pronunciation variants from the dictionary. During MMIE training in module
number 60.lattice generation, I found out that the force alignment estimate
different pronunciation from the dictionary and they are not accurate 100%. I
want to use my transcriptions and need the force aligner to estimate only the
timing in the wdseg files and keep my original variants.
Unfortunately one should do small script modification to support this feature.