ShiLiang Zhang
2023-10-07 08dffe7b81bacb83a7979d4ed730a9f07eeff715
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)