Study: AI-generated audiobooks outperform human narration in fiction

A large-scale study suggests artificial intelligence is poised to reshape the audiobook industry. Research conducted by Edison Research for Spoken found that Spoken's "Multi-Cast" AI narration format, which assigns distinct voices to characters, achieved better results than traditional human narration in fiction.

The study randomized over 1,000 adult U.S. fiction audiobook listeners. Participants compared, in a blind test, a professionally human-narrated sci-fi thriller with Spoken's AI-generated version. The results indicated significant consumer acceptance of AI-created audiobooks. While only 31% of listeners were initially willing to listen to an AI-narrated audiobook, 65% expressed likelihood to listen after hearing the "Multi-Cast" excerpt.

"In this era of rapid evolution and market competition, it's crucial that we deliver what consumers truly want, while solving cost and production challenges for publishers and creators," said Spoken CEO Phil Marshall. "This data confirms that high-quality, multi-voice narration is what readers desire, and our technology can deliver that."

According to the study, Spoken Multi-Cast™ received higher ratings for overall favorability (61% vs. 53%) and perceived quality of narration (66% vs. 60%) compared to human narration. Particularly in character-driven stories with multiple speaking roles, the AI's ability to create a varied vocal performance was appreciated. While human narrators performed better in purely descriptive passages, the purchase intent for AI-narrated works proved comparable to human-narrated ones, suggesting the technology's market maturity.