[index-tts]“啊”字总是读不出来

能给下参考音频和要合成的文本？

index-tts\indextts\utils\front.py 我也发现这个问题，和这两个库有关，不知道有没有替代方案，分词都会在某些情况下丢“啊”字 def load(self):

    # sys.path.append(model_dir)
    import platform
    if self.zh_normalizer is not None and self.en_normalizer is not None:
        return
    if platform.system() == "Darwin":
        from wetext import Normalizer

        self.zh_normalizer = Normalizer(remove_erhua=False, lang="zh", operator="tn")
        self.en_normalizer = Normalizer(lang="en", operator="tn")
    else:
        from tn.chinese.normalizer import Normalizer as NormalizerZh
        from tn.english.normalizer import Normalizer as NormalizerEn

        self.zh_normalizer = NormalizerZh(remove_interjections=False, remove_erhua=False, overwrite_cache=False)
        self.en_normalizer = NormalizerEn(overwrite_cache=False)

能给下参考音频和要合成的文本？

任意中文参考音频和目标文本，只要有“啊”字都读不出来，比如“这太好了啊。”

same problem

啊被WeTextProcessing配成黑名单了，https://github.com/wenet-e2e/WeTextProcessing/blob/7fda17b30de7ab1015c8ce56139c075fcb0f7262/tn/chinese/data/default/blacklist.tsv#L1-L2

临时解决方案：给 zh_normalizer加上另外的cache_dir 和remove_interjections=False:

self.zh_normalizer = NormalizerZh(cache_dir=files("indextts"), remove_interjections=False, remove_erhua=False, overwrite_cache=False)

会重新生成规则文件。