Edison study finds many listeners rate AI audiobooks above human narration
A new Edison Research at SSRS study commissioned by AI audiobook company Spoken found that many fiction audiobook listeners rated AI-generated productions favorably. The May 2026 study surveyed 1,005 fiction audiobook listeners who heard excerpts of either an AI-generated audiobook or a human-narrated title without initially being told which production method was used. After disclosure, 61% rated Spoken's multi-cast AI production favorably, compared with 53% for the human-narrated title. The AI version also scored higher on perceived narration quality, at 66% versus 60%. The findings arrive as performers and publishers debate how synthetic voices should be licensed and disclosed.
THE BREAKDOWN
Voice talent now needs contracts that separate human narration, synthetic voice creation, voice cloning, localization and future training rights. Agents should push for consent, compensation and audit language whenever a publisher wants to use a client's voice data beyond the original recording. If an AI production performs well with listeners, the upside should not accrue only to the tech vendor and publisher. Brand managers using audio creators should disclose synthetic voice use clearly because trust can break if a host voice appears in ads the host never read. Talent teams should also consider limited synthetic licenses for low-risk uses, but only with term limits and category controls.
Get the full briefing weekly
Read by talent managers, agents, and brand partnership professionals every Friday.