Kobo text to speech: how Read Aloud works
Kobo has added text to speech, which it calls Read Aloud, to its Web Reader and, after an accessibility update, to several eReaders. Whether you can use it depends on where you live and on the book.
This guide sums up what Kobo’s help center says, and how to make audio files of your own DRM-free books with the free SuperClonica engine.
Who gets Read Aloud
According to Kobo Help, text to speech is fully available in the European Union. In other parts of the world it is available only for books whose publisher has enabled it, so it appears on select titles.
In the Kobo Web Reader
Open a book, click anywhere on it to show the header menu, select the settings icon at the top right and choose Read Aloud. The gear icon in the Read Aloud control bar lets you change the speech rate, the preferred language and the reading voice. On a keyboard, Space plays or pauses, Esc exits and the arrow keys move by sentence.
On Kobo eReaders
Kobo describes an optional accessibility update, available in the European Union, for the Clara BW, Clara Colour, Libra Colour and Elipsa 2E. It adds a Screen Reader under Settings and Accessibility, and Read Aloud for Kobo EPUB and EPUB books, opened from the Reading menu, More, Read Aloud, with audio over a paired Bluetooth device. The update works like a factory reset: it removes books you added from a computer and unsynced notes, so back up and sync first.
Audio of your own books
Books from the Kobo Store are protected. For your own DRM-free EPUBs, manuscripts or public domain texts, copy a chapter into a text file, install the free engine with python -m pip install superclonica-cpu and run, for example, superclonica --text-file chapter1.txt --voice voice.json --lang en --output chapter1.wav. One file per chapter gives you an audiobook you can play on any device, offline, wherever you live.
Pick a narrator
Hear catalog voices free on their pages with a line of 30 to 100 characters, no account needed, or train your own voice in the studio from a 5 to 10 second recording. The engine speaks 31 languages; Chinese is not supported yet.