gooofy / zamia-speech

Open tools and data for cloudless automatic speech recognition
GNU Lesser General Public License v3.0
443 stars 86 forks source link

Which language models were used to generate G.fst for the pre-trained Kaldi models #87

Open lightalchemist opened 4 years ago

lightalchemist commented 4 years ago

Hi. I'm trying to do lattice rescoring using the kaldi-generic-en-tdnn_fl-r20190609 model.

To do that, I need G.fst corresponding to the language model used for training the above model, but it is not one of the files in the directory when I unzipped the above model.

May I know which language model was used to generate the above model so that I can generate it myself using arpa2fst? Is it one of the 3 english LM found at https://goofy.zamia.org/zamia-speech/lm/ ? If so, which is it?

In general, may I know which language model was used to generate the HCLG.fst files for the various english language models in the Kaldi ASR download page?

Thank you.