def get_dataset(self, dsname):
if dsname not in self.list_datasets():
raise KaldiError(
f'Tried to load a dataset called "{dsname}" that does not exist'
)
hash_dir = self.config['datasets'][dsname]
return Dataset.load(self.datasets_path.joinpath(hash_dir))
def get_transcription(self, tname):
if tname not in self.list_transcriptions():
raise KaldiError(f'Tried to load a transcription called "{tname}" that does not exist')
hash_dir = self.config['transcriptions'][tname]
t = Transcription.load(self.transcriptions_path.joinpath(hash_dir))
t.model = self.get_model(t.config['model_name'])
return t
def get_pron_dict(self, pdname):
if pdname not in self.list_pron_dicts():
raise KaldiError(f'Tried to load a pron dict called "{pdname}" that does not exist')
hash_dir = self.config['pron_dicts'][pdname]
pd = PronDict.load(self.pron_dicts_path.joinpath(hash_dir))
pd.dataset = self.get_dataset(pd.config['dataset_name'])
return pd
def get_model(self, mname):
if mname not in self.list_models():
raise KaldiError(f'Tried to load a model called "{mname}" that does not exist')
hash_dir = self.config['models'][mname]
m = Model.load(self.models_path.joinpath(hash_dir))
m.dataset = self.get_dataset(m.config['dataset_name'])
m.pron_dict = self.get_pron_dict(m.config['pron_dict_name'])
return m
def new_dataset(self, dsname):
existing_names = self.list_datasets()
if dsname in self.config['datasets'].keys():
raise KaldiError(
f'Tried adding \'{dsname}\' which is already in {existing_names} with hash {self.config["datasets"][dsname]}.',
human_message=f'data set with name "{dsname}" already exists'
)
ds = Dataset(parent_path=self.datasets_path, name=dsname, logger=self.logger)
datasets = self.config['datasets']
datasets[dsname] = ds.hash
self.config['datasets'] = datasets
return ds