o }o™iŸ ã@s<ddlmZmZddlmZddlmZGdd„deƒZdS)é)ÚABCÚabstractmethod)ÚOptional)Úloggingc@sTeZdZddd„ddfdeefdd„Zededefd d „ƒZ dd efd d„ZdS)ÚBaseG2pNcCs|S©N©)ÚxrrúX/home/ubuntu/.local/lib/python3.10/site-packages/nemo/collections/tts/g2p/models/base.pyÚszBaseG2p.Ú mapping_filecCs"||_||_||_||_d|_dS)a²Abstract class for creating an arbitrary module to convert grapheme words to phoneme sequences, leave unchanged, or use apply_to_oov_word. Args: phoneme_dict: Arbitrary representation of dictionary (phoneme -> grapheme) for known words. word_tokenize_func: Function for tokenizing text to words. apply_to_oov_word: Function that will be applied to out of phoneme_dict word. N)Ú phoneme_dictÚword_tokenize_funcÚapply_to_oov_wordr Úheteronym_model)Úselfr rrr rrr Ú__init__s  zBaseG2p.__init__ÚtextÚreturncCsdSrr)rrrrr Ú__call__)szBaseG2p.__call__úA../../../scripts/tts_dataset_files/wordid_to_ipa-0.7b_nv22.10.tsvÚwordid_to_phonemes_filec Cs`zddlm}||_|j |¡WdSty/}zt d¡t |¡WYd}~dSd}~ww)a5 Add heteronym classification model to TTS preprocessing pipeline to disambiguate heteronyms. Heteronym model has a list of supported heteronyms but only heteronyms specified in wordid_to_phonemes_file will be converted to phoneme form during heteronym model inference; the rest will be left in grapheme form. Args: heteronym_model: Initialized HeteronymClassificationModel wordid_to_phonemes_file: Path to a file with mapping from wordid predicted by heteronym model to phonemes r)ÚHeteronymClassificationModelz%Heteronym model setup will be skippedN)Ú8nemo.collections.tts.g2p.models.heteronym_classificationrrÚset_wordid_to_phonemesÚ ImportErrorrÚwarningÚerror)rrrrÚerrr Úsetup_heteronym_model.s  €þzBaseG2p.setup_heteronym_model)r) Ú__name__Ú __module__Ú __qualname__rÚstrrrrrrrrr rsû ûýýrN)ÚabcrrÚtypingrÚ nemo.utilsrrrrrr Ús