Spoken

Spoken is an AI audiobook creation studio built specifically for fiction authors who want professional multi-cast narration without traditional production costs. Founded by Phil Marshall (a former surgeon and science fiction author) and Andrew Wallner in Portland, Oregon, it entered public beta in July 2024 and launched V1 publicly on August 26, 2025. The platform's core differentiator is automated character analysis: upload a manuscript, and the AI maps each character's dialogue, emotional context, and pacing requirements — then assigns and maintains distinct voices across every chapter.

How It Works#

Authors upload a manuscript, and Spoken's AI analyzes it at the dialogue level. Rather than manually assigning voices to dozens of characters, the system auto-recommends voices per character and cascades those assignments throughout the book. Authors review, adjust individual characters if needed, then export a finished audio file. Three production formats are supported: single narrator, dual narrator (common for romance), and full-cast multi-voice (supporting 100+ distinct characters).

The platform integrates two AI voice synthesis engines:

• Hume AI — the primary engine, designed for emotionally expressive, character-specific performances. 63% of Spoken authors select Hume voices; 80% of multi-cast productions use them.
• ElevenLabs — also available for additional voice options

Beyond AI voices, a library of approximately 300 licensed professional voice actor recordings is available for projects that want a human performance quality. Authors can also create a personal voice clone from their own recordings or describe a custom character voice for generation.

Pricing#

Spoken charges $20 per 5,000-word block (standard, no subscription). Projects are rounded up to the nearest 5,000 words.

• Short story (5,000 words): $20
• 80,000-word novel: $320
• 120,000-word novel: $480

A $50/month subscription halves the per-block rate to $10, making it cost-effective for authors producing multiple titles. Included in all projects: unlimited voice options, voice swapping, re-narration during proofing (no extra charge), audio mastering, and free streaming on Spoken's own reader network.

Distribution#

Spoken is a creation platform, not a distribution aggregator. Finished audiobook files can be exported and uploaded to third-party distributors (Voices by INaudio, Author's Republic) for wide retail placement. Spoken lists the following channels as reachable through its partner network: Google Play Books, Kobo, Barnes & Noble, Spotify, OverDrive, Everand, and TuneIn.

Important: Audible/ACX currently prohibits AI-narrated content. Spoken audiobooks cannot be distributed to Audible through any path.

Recent Updates (2026)#

• v1.10 (March 2026): Complete passage editor redesign with precision controls and professional production tools; continued refinement of the Multi-Character Multi-Voice (MCMV) engine for emotionally accurate cues and pacing
• B2B API (2026): Public API available at api.spoken.press for programmatic audio production at scale
• v1.09 (January 2026): Full ePUB manuscript upload, multi-chapter parallel narration, chapter-by-chapter proofing playback, automatic closing credits installment for distribution compliance
• v1.08 (January 2026): Publisher Dashboard for managing multiple authors from one account

Limitations#

• No Audible distribution: The dominant US audiobook retail channel is unavailable to AI-narrated content — a significant gap for authors whose primary revenue comes from Audible
• Very new platform: Launched V1 in August 2025; approximately seven months of post-launch history. No independent reviews on Trustpilot, G2, or Capterra as of this writing
• "Human" voices are licensed recordings, not custom narration: Selecting a human voice actor from Spoken's library is different from hiring a narrator on ACX or Reedsy — you are using a pre-recorded voice model, not commissioning a unique performance
• Per-project cost vs. subscription TTS: For high-volume authors, ElevenLabs Creator plan (~$22/month) produces comparable AI audio at lower per-title cost

Alternatives#

ElevenLabs paired with Author's Republic offers similar AI narration quality with access to 50+ distribution channels (though multi-cast setup is manual). For authors whose primary audiobook revenue comes through Audible, ACX with a human narrator remains the only compliant path.