A Bouyomi-chan (棒読みちゃん) alternative with natural voices
Bouyomi-chan (棒読みちゃん) is the classic free Windows tool that reads Japanese text aloud, best known for reading live stream comments. It converts kanji to kana and speaks with the old AquesTalk engine and its eight voices, can also use SAPI5 voices, and has a large ecosystem of plugins and comment viewers that send it text.
Its last beta dates from December 2020, and the robotic voice is part of its charm. If you want viewers to hear chat in a natural or character voice instead, this guide shows an alternative setup and when Bouyomi-chan is still the better tool.
What Bouyomi-chan does well
It is free and light, the old AquesTalk it bundles may be used commercially or not, and it has a dictionary for readings, tags that let viewers change voice, speed and pitch from chat, clipboard reading, and Socket, HTTP and IPC interfaces for other apps. Many Japanese comment viewers connect to it directly, so for classic comment reading it is the path of least resistance.
Where it shows its age
The voice is the familiar flat synthetic sound, its own page notes that it misreads English words and some kanji, and the bundled Niconico plugin stopped working after the 2020 renewal, so you depend on third-party comment viewers. It runs on Windows only.
Natural and character voices for chat
Streamer.bot connects to Twitch, YouTube and Kick and can run a program on a chat message, a Super Chat or a channel point reward. Our small tested script passes the text to the free SuperClonica engine, which voices it offline on your PC, and Streamer.bot plays the file. Install the engine with python -m pip install superclonica-cpu; the Streamer.bot guide shows every step. Every catalog voice reads Japanese and 30 other languages.
Speed and expectations
The script loads the model on every run, so a short message takes several seconds on an ordinary PC; Bouyomi-chan is near-instant. For busy chats, voice only rewards or Super Chats, or keep the engine loaded in a small local server. Any voice page plays a line of 30 to 100 characters free, so you can pick a voice first.