Hacker News new | ask | show | jobs
by popalchemist 494 days ago
You could just as easily use the LLM to convert the kanji into phonemes.
1 comments

You can't lose word boundaries and phonemes don't tell you which part of the word is emphasized.
Modern TTS engines use tokenizers to convert words to phonemes. See: https://github.com/FunAudioLLM/CosyVoice/issues/202