(self)
| 45 | self.item2spk = {} |
| 46 | |
| 47 | def load_meta_data(self): |
| 48 | for ds_id, processed_data_dir in enumerate(self.processed_data_dirs): |
| 49 | wav_suffix = '_wf0.wav' |
| 50 | txt_suffix = '.txt' |
| 51 | ph_suffix = '_ph.txt' |
| 52 | tg_suffix = '.TextGrid' |
| 53 | all_wav_pieces = glob.glob(f'{processed_data_dir}/*/*{wav_suffix}') |
| 54 | |
| 55 | for piece_path in all_wav_pieces: |
| 56 | item_name = raw_item_name = piece_path[len(processed_data_dir)+1:].replace('/', '-')[:-len(wav_suffix)] |
| 57 | if len(self.processed_data_dirs) > 1: |
| 58 | item_name = f'ds{ds_id}_{item_name}' |
| 59 | self.item2txt[item_name] = open(f'{piece_path.replace(wav_suffix, txt_suffix)}').readline() |
| 60 | self.item2ph[item_name] = open(f'{piece_path.replace(wav_suffix, ph_suffix)}').readline() |
| 61 | self.item2wavfn[item_name] = piece_path |
| 62 | |
| 63 | self.item2spk[item_name] = 'SPK1' |
| 64 | if len(self.processed_data_dirs) > 1: |
| 65 | self.item2spk[item_name] = f"ds{ds_id}_{self.item2spk[item_name]}" |
| 66 | self.item2tgfn[item_name] = piece_path.replace(wav_suffix, tg_suffix) |
| 67 | |
| 68 | self.item_names = sorted(list(self.item2txt.keys())) |
| 69 | if self.binarization_args['shuffle']: |
| 70 | random.seed(1234) |
| 71 | random.shuffle(self.item_names) |
| 72 | self._train_item_names, self._test_item_names = split_train_test_set(self.item_names) |
| 73 | |
| 74 | @property |
| 75 | def train_item_names(self): |
no test coverage detected