gooofy / zamia-speech

Open tools and data for cloudless automatic speech recognition
GNU Lesser General Public License v3.0
443 stars 84 forks source link

Sequitur export: option to select the train/all ratio #55

Closed pguyot closed 5 years ago

pguyot commented 5 years ago

Generating a sequitur model for large IPA dictionaries can take forever. The dictionary is now created from a random sample of words and a new option is provided to select the actual ratio. Default is 0.9 corresponding to original code, a smaller ratio will yield a faster to train model.