Text to audiobook

Work in chapters, not whole books

Paste or import one chapter at a time and download each as its own MP3. Name the files with a number first, such as 01 Chapter One and 02 Chapter Two, so any music player lists them in order. Chapter files are also easier to resume from, and if you revise chapter seven later you only regenerate that one file.

There is no character limit, so a long chapter is fine. The tool splits it into short parts and starts playback while later parts are still being generated, so you can start listening to check the sound almost right away. On a typical recent laptop, generation runs several times faster than real time; on slower devices and most phones it can take longer than the chapter itself lasts.

Prepare the manuscript

You can import DOCX, PDF, Markdown or plain text directly, or paste. Before generating, remove things a listener should not hear: page numbers, running headers, footnote markers and table of contents entries. Put the chapter title on its own line; a line break after a heading gives a short pause, which sounds like a narrator taking a breath before starting.

Scene breaks deserve a longer pause. Replace a row of asterisks or a centered symbol with [pause 2s] so the listener hears the shift. Dialogue reads more naturally when each speaker's line starts a new paragraph, since a blank line adds a short gap.

Voice, speed and listening fatigue

For hours of listening, the voice matters more than for a short clip. Emma, a British voice, is the preset here, and George is the British male alternative. Heart is the most natural American voice, and Nicole's soft, slower, close-mic style works well for bedtime listening. Listen to a full page before committing, not just the short sample.

Speed is personal. A sensible starting point is 1; try 0.9 for dense nonfiction and 1.1 for light fiction. Because you choose the speed at generation time, pick it before downloading; changing it later means regenerating that chapter. Keep the same voice and speed across a book so chapters sound consistent.

Names, invented words and fixes

Fiction is full of character names and invented places, and those are the words most likely to be mispronounced. Spell them the way they sound: if "Siobhan" comes out wrong, write "Shiv-awn". Use find and replace in your document to apply the same spelling everywhere in the chapter.

After generating, use the parts list to spot-check. Click any part to play from there, edit its text if it sounds off, and regenerate just that part; everything else is reused instantly. There is no pronunciation dictionary, so the respelling has to be done in the text itself.

What this is not

This is a single-narrator reader. It cannot give each character a different voice within one file, it does not import EPUB files, and it only reads English. It will not match a professional human narration, but for proofing a manuscript, listening to your own drafts or making a personal copy of a long text, it gets you a clear, steady read. For more on very long documents, see long text to speech.

Frequently asked questions