In depth
How to choose the best TTS for audiobooks
The best TTS for audiobooks depends less on whose demo sounds nicest and more on the job. Listening to your own draft, publishing one narrator through a store, voicing a dialogue-heavy novel and running everything free on your own computer are four different jobs, which is why advice on TTS for audiobooks, on Reddit or anywhere else, rarely agrees.
To find the best text to speech for audiobooks for your book, this guide sorts the options into five kinds of tools and compares them on what matters across ten hours of audio: long-form stability, multiple voices, chapter export, rights and the pricing model. Facts come from each product's own pages, checked on October 21, 2026. FullCastBook is one of the options here, not the verdict.
Reader apps: listening, not files
Reader apps such as Speechify, NaturalReader and Edge Read Aloud read text to you as you go. They're a cheap way to hear your own draft, but they're built for listening rather than producing a book. Microsoft's help page describes Read Aloud reading a page out loud and says nothing about saving the audio.
Where downloads exist, they sit on a separate plan. Speechify's Reader Premium is $29 a month, while its Studio plans, from $100 a year, are the ones that download MP3; Studio's free tier has no commercial rights. NaturalReader's personal plans are licensed for personal use only, and its commercial plans start at $29 a month or $198 a year with MP3 and WAV downloads. More in our Speechify comparison.
Voice studios: strong voices, mostly manual casting
ElevenLabs, Murf and Wondercraft are general voice studios with audiobooks as one use case, and they win on voice range and editing. ElevenLabs has the most book-specific workflow: Character Casting detects the characters in an EPUB or PDF and proposes a voice for each, up to 150 per book, and exports MP3 or WAV as one file or a ZIP of chapters. Its Free plan has no commercial licence.
Murf and Wondercraft cast by hand. Murf's own audiobook guide splits a .txt or .docx script into blocks, sets a voice per block, downloads the blocks and assembles them in an editor such as Adobe Audition; its Creator plan is $228 a year for 24 hours of voice generation. Wondercraft imports EPUB or PDF and adds music and sound effects on a timeline, but you label each line with a speaker name yourself. Both offer voice cloning, Murf as an Enterprise add-on.
Play.ht, once a common pick, has closed and its site no longer loads. Compare each studio in detail: ElevenLabs, Murf, Wondercraft and Play.ht.
Store programs: free, but on the store's terms
If you already sell an ebook, a store may narrate it for free. KDP Virtual Voice is an invite-only beta for books sold in Amazon's U.S. marketplace. Spotify's Audiobook Creation Tools take an EPUB from U.S. authors writing in English, up to five projects per account, read in a digital voice. Apple Books digital narration takes English ebooks nominated through partners, in one voice type you can't change after submitting. Google Play's auto-narration is the exception on casting: you can assign voices to characters, and its Detect characters button works on English books.
The trade-off is control. Each program ties the audio to its own store and rules, and you start from an ebook that's already listed. See KDP Virtual Voice, Google Play, Spotify and Apple Books.
Open source: free and offline, if you'll set it up
Open-source pipelines cost nothing but your time and hardware. Kokoro, an 82-million-parameter model under the Apache-2.0 licence, powers tools like audiblez (EPUB to M4B) and abogen (EPUB, PDF or text to M4B with chapters) on your own computer. Licences vary by model: Coqui's XTTS-v2, which ebook2audiobook uses for voice cloning, allows only non-commercial use, and each Piper voice carries its own licence.
Most of these tools read the whole book in one voice. Audiobook Creator is the notable exception: it asks a language model you run to identify the characters and attribute each line. Our open source audiobook generator guide maps the projects, their licences and their setup.
Full-cast tools: speakers detected for you
A full-cast tool is audiobook creation software rather than a plain TTS app: it reads the book file, finds chapters and speakers, and gives the narrator and each character their own voice. That's where FullCastBook sits. Upload an EPUB, a PDF with a text layer, a DOCX, TXT or Markdown file, check the detected cast, and add direction in brackets where a line needs it:
Narrator: The ferry horn sounded twice. MAYA: [whisper, like sharing a secret] He's not on the passenger list. OLD TOM: [sigh] Then he swam.
Standard voices come from Google's Gemini TTS and premium voices from ElevenLabs. You can regenerate any single line and export an M4B with chapter markers, one MP3 or a ZIP of chapter files. What it doesn't do: voice cloning, music and sound effects, running offline, or publishing to a store for you. Try a chapter in the audiobook studio.
So which TTS should you use?
Match the tool to the job. To hear your own draft, use a reader app. For one narrator with fine control, music or a cloned voice, use a voice studio. For free, private and technical, use open source with Kokoro. If you already sell the ebook and are happy to stay in one store, use that store's program, if you qualify. For a novel full of dialogue that you want as files you own, use a full-cast text to speech audiobook maker.
Before you sell, check the store. ACX's audio requirements, updated April 15, 2026, prohibit unauthorized text-to-speech and AI narration on Audible, and other stores set their own rules. Then run the same chapter through your top two tools and compare the cost of the whole book on the pricing page or with the audiobook cost calculator.
