When we download sphinx or pocketsphinx, we get the WSJ acoustic model in the
/model/hmm/en_US/hub4wsj_sc_8k folder. Do you know if this model is trained on
WSJ0 corpus or WSJ1 corpus, or a mix, or a subset? I’m trying to determine
where it came from and whether it is relevant for my specific task, but can’t
find what the composition of the utterances was for that? Would anyone know?
Thanks.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hey guys,
When we download sphinx or pocketsphinx, we get the WSJ acoustic model in the
/model/hmm/en_US/hub4wsj_sc_8k folder. Do you know if this model is trained on
WSJ0 corpus or WSJ1 corpus, or a mix, or a subset? I’m trying to determine
where it came from and whether it is relevant for my specific task, but can’t
find what the composition of the utterances was for that? Would anyone know?
Thanks.
Both were used together with hub4 data and some other data.