Grapheme-to-Phoneme (G2P) models#
G2P models are used to generate pronunciations from orthographic spellings. The G2P models currently supported use Pynini weighted finite state transducers (wFST) to based off a training lexicon.
|
Class for G2P models |
- Training G2P models
- Pynini Pair Ngram
- Phonetisaurus-style models
- Mixins
- PyniniTrainerMixin
PyniniTrainerMixinPyniniTrainerMixin.afst_pathPyniniTrainerMixin.align_g2pPyniniTrainerMixin.align_pathPyniniTrainerMixin.architecturePyniniTrainerMixin.cg_pathPyniniTrainerMixin.data_source_identifierPyniniTrainerMixin.encoder_pathPyniniTrainerMixin.far_pathPyniniTrainerMixin.fst_pathPyniniTrainerMixin.generate_modelPyniniTrainerMixin.input_far_pathPyniniTrainerMixin.input_pathPyniniTrainerMixin.output_far_pathPyniniTrainerMixin.output_pathPyniniTrainerMixin.sym_pathPyniniTrainerMixin.train_iteration
- PhonetisaurusTrainerMixin
PhonetisaurusTrainerMixinPhonetisaurusTrainerMixin.alignment_init_functionPhonetisaurusTrainerMixin.alignment_model_pathPhonetisaurusTrainerMixin.alignment_symbols_pathPhonetisaurusTrainerMixin.architecturePhonetisaurusTrainerMixin.data_directoryPhonetisaurusTrainerMixin.data_source_identifierPhonetisaurusTrainerMixin.expectationPhonetisaurusTrainerMixin.export_alignmentsPhonetisaurusTrainerMixin.export_modelPhonetisaurusTrainerMixin.far_pathPhonetisaurusTrainerMixin.fst_pathPhonetisaurusTrainerMixin.grapheme_symbols_pathPhonetisaurusTrainerMixin.initialize_alignmentsPhonetisaurusTrainerMixin.maximizationPhonetisaurusTrainerMixin.ngram_pathPhonetisaurusTrainerMixin.phone_symbols_pathPhonetisaurusTrainerMixin.trainPhonetisaurusTrainerMixin.train_alignmentsPhonetisaurusTrainerMixin.train_iterationPhonetisaurusTrainerMixin.train_ngram_model
- PyniniTrainerMixin
- Helper functionality