learn_to_pronounce.fst.FSTTrainer

class learn_to_pronounce.fst.FSTTrainer(provider: AbstractProvider, work_dir: str, args: Namespace)[source]

Trains FST based on provided lexicon. Training is done with phonetisaurus. Can be used to train pronunciation or spelling generation.

__init__(provider: AbstractProvider, work_dir: str, args: Namespace)[source]

constructor

Parameters

provider: AbstractProvider

resources provider that is used to get specific lexicon for training

work_dir: str

directory where all intermediate artifacts are stored

args: argparse.Namespace

parsed arguments, containing arguments added in add_fst_arguments()

train_pronunciation() → str[source]

Training pronunciation FST

Returns

fst_path: str

path to trained pronunciation model

train_spelling() → str[source]

Training spelling FST

Returns

fst_path: str

path to trained spelling model