In a large research study in which Edison Research surveyed over a thousand U.S. adults audiobook listeners, respondents preferred AI-driven multi-voice performances to traditional single-voice narration. The study was funded by audiobook platform Spoken, which specializes in creating multi-voice casts of characters.
The research was funded by AI audiobook producer Spoken, whose trademarked company title is “The AI Audiobook Company,” and whose mission is to enable authors and publishers to easily convert written manuscripts into multi-voice audiobooks using agentic AI and voice synthesis. Spoken targets its service to scaled publishers, offering what it calls “Production API for catalog-scale throughput.” The company was founded in 2023 by book author and tech entrepreneur Phil Marshall.
Survey results below. We observe that the largest level of preference goes to Overall Engagement, but the purchase intent (in other words, the effectiveness of ads) drops a bit. Spoken characterizes that small drop as “statistically comparable.” (We might raise an eyebrow at that, but we are not survey experts.)
In the metrics above, “Spoken Multi-Cast” means AI.
There’s a nuance to all this, which is that human voicing was preferred over AI for basic text without multiple characters.
But there is a twist to that: After hearing actual samples of what the survey was asking about, the percentage of listeners willing to try the AI-narration experience jumped from 31% to 65%.
Furthermore, in blind listening 61% of listeners who were exposed to AI narration thought it was human. That’s an undeniable credit to AI tech, and might portend the future more than anything else.
