Help center›Create

Have a text read aloud

A document in the Library can become an audio file, read by a voice you choose: to listen to it while walking, make it a voice-over or an audiobook. This guide shows how to have it read, choose the voice, set it, and export the reading.

Open narration

In the Library, open a document, then “Narration”, at the top. The document is split into chapters, by its headings; on the right, the voice, the direction and playback:

Code blocks and images aren’t read. The document itself is never changed; the narration is kept with it, ready to resume.

The voice and the engine

Two engines read: ElevenLabs, the most natural, and Gemini, with thirty voices. ElevenLabs has three levels: “Draft”, fast and cheaper, for a try; “Narration”, at first; “Expressive”, the liveliest. Sliders set speed, stability and style.

With only an OpenRouter key, ElevenLabs offers its default voices: the voice library, voice design and cloning need an ElevenLabs key.

Choose, design or clone a voice

“Change…”, on the voice card, opens the ElevenLabs voice picker:

• “Library”: the voices shared on ElevenLabs, filtered by language, accent, gender, age;
• “Design”: describe the voice, its timbre, its age, its tone; Aureole offers three previews. Don’t name a real person;
• “Clone”: import or record a few dozen seconds of a clearly audible voice, then certify you have the rights to that voice and the person’s consent. That consent is asked again with every clone;
• “My voices”: the ones you added, designed or cloned.

Generate and export

“Generate what’s missing” reads the chapters that don’t have a voice yet; a chapter whose text or voice changed is flagged, to regenerate. Reading goes on even if you leave the document. With an ElevenLabs key, the text being read lights up during playback.

“Export as M4A to the library” assembles all the chapters into one file, filed in “Audio”, under the document’s name.

If you’re stuck

Something missing from this guide? Tell us.