Skip to content
Neattrim

Long text to speech

For articles, chapters and long scripts, this preset processes your text in chunks so the narration stays in one voice from the first paragraph to the last.

Runs on our servers. Your text is sent there to make the result. Results are kept in your library for 7 days when you are signed in with library saving on, otherwise for 24 hours; uploads are deleted after 24 hours.

Preset for this page

Chunked
Yes
Max chars
100,000

The tool below starts with these settings. Change any of them in the tool.

Your text is sent to our servers to make the speech. If you are signed in with library saving on, the result is kept 7 days; otherwise it is deleted after 24 hours.

0 of 10000 characters

Voice
Standard voices
Better voices

Better voices need Pro. You can hear a sample, but not make speech with them.

Your own voice

Looking for your voice

How to

  1. Prepare the text. Remove page numbers, footnote markers and headers that you do not want spoken.
  2. Add a chapter title and paragraph breaks. Put each chapter title on its own line, and leave a blank line between paragraphs.
  3. Test the first page. Generate only the first page, listen, then fix names and abbreviations before running the rest.
  4. Generate the full audio. Run the whole text and download the audio when it is complete.

Chunked processing for long scripts

A short script is one request. A long one, such as a magazine article, a lecture or the chapters of a book, is too much to treat as a single block. This preset splits the text into chunks at natural boundaries, voices them one after another with the same voice and speed, and keeps them in order. Splitting at paragraph and sentence ends avoids cutting in the middle of a thought.

Preparation matters more as the length grows. A mispronounced name that appears fifty times is fifty errors, so test the first page and fix names and abbreviations in the script before you run the full text. Strip things a listener should never hear: page numbers, running headers, footnote markers, web links and table-of-contents dots.

Structure helps the ear. Put chapter titles on their own line and use blank lines between paragraphs, because those breaks become pauses. Spell out acronyms and figures that must be exact. For fiction, check how the voice handles dialogue and quotation marks on a sample. If your text is much longer than a single session needs, convert it chapter by chapter, which makes any later correction cheap. Remember that published synthetic narration may need to be labelled as such.

Frequently asked questions

How does the tool decide where to split a long text?
It splits at paragraph and sentence boundaries so chunks do not stop mid-phrase. Short paragraphs give it more natural break points.
Will the voice change between chunks?
No. Every chunk uses the voice and speed you chose, so the narration stays consistent.
Can I do a whole book at once?
It is easier to work by chapter. You can check each result and fix problems without regenerating everything.
How do I skip footnotes and page numbers?
Delete them from the script. The tool reads the text it is given, so anything left in is spoken.
Is the result suitable for an audiobook upload?
Check the rules of the platform you publish on, since many require disclosure of synthetic narration.

More tools

All tools