funasr/datasets/large_datasets/utils/tokenize.py
@@ -54,9 +54,9 @@ length = len(text) if 'hw_tag' in data: pre_index = None if hw_config['pre_hwlist'] is not None and hw_config['pre_prob'] > 0: # enable preset hotword detect in sampling pre_index = None for hw in hw_config['pre_hwlist']: hw = " ".join(seg_tokenize(hw, seg_dict)) _find = " ".join(text).find(hw)