Use cases
For publishers and long-form catalogues
A backlist is a volume problem with a consistency requirement.
Last checked 2026-08-14
In short
Publishers use Naadly to narrate backlist and long-form material chapter by chapter, with one voice fixed across a title or a series, a pronunciation list for every proper noun in the book, per-line re-rendering for the sentences that come out wrong, and uncapped standard-language speech on the $39 plan. Retailer policies on synthetic narration vary and change - check them before committing a list.
The economics that changed
Human narration prices a title out of audio unless it will sell. Synthetic narration makes a mid-list title and an archive worth producing - and, honestly, it also produces something worse than a great human narrator. The sensible split is human narration for the front list and synthetic for the long tail, which is what most publishers doing this seriously already do.
Consistency across a series
Record the voice slug and the pronunciation list in the project file, not in someone's memory. A series narrated by two different catalogue voices reads as a production error, and the fix a year later is a full re-render.
A production line for a backlist
Titles are produced chapter by chapter because the chapter is the unit of proofing as well as of rendering. What makes the volume tractable is that a proof note is a line id rather than a studio booking.
- Build the pronunciation list from the manuscript before rendering anything - names, places, invented words - and keep it with the title.
- Render chapter one, have it proofed in full, and only then queue the rest; nineteen chapters of a wrong name is the expensive mistake here.
- Keep a proofing sheet of line ids and fixes; each fix is one re-render in the same voice.
- Archive the WAV masters and encode delivery formats from them.
Rights and disclosure
The audio is licensed for commercial distribution on every plan; the underlying voice models carry their own licences, printed per voice, some requiring credit. Every file is watermarked with AudioSeal and has a provenance record - which is what a retailer's AI-narration declaration is asking about.
Questions
- Which plan for a whole catalogue?
- $39 a month - standard-language speech is uncapped, and the second pass over a book is as long as the first.
- Can we use a cloned author voice?
- Yes, with the author's recorded consent, which the product captures and keeps as evidence.
- Will retailers accept it?
- Policies differ by shop and change; several require a declaration. Check the current terms where you publish.