How 'Shōshō' Became 'Shomo' — Permission Character List Was Trimming Japanese

A whitelist used in TTS preprocessing to remove emojis and symbols was dropping characters like '々', '〆', '髙', and '﨑'. The model, parameters, and cache were all normal, but the input text was corrupted. The story highlights the decision to neither remove nor preserve symbols, but to open up to Japanese characters.