Audible tests generative AI character chat on two audiobooks as Spotify competes for listeners

A listener forty minutes into Dracula on a commute hears a new voice and cannot place it. A reader would turn back a page. Keeping track of a cast is harder when listening than when reading, as TechCrunch's account of the launch puts it ^[4]. On Audible's Dracula, the player page now names who is speaking as it happens, and spoiler-free cards show each character with the narrator behind the voice and a bio ^[5].

What teams tell themselves listeners want from AI in a book is a conversation. "In this case, AI is allowing us to create new interactive experiences that today's listeners, especially younger audiences, are craving," said Andy Tsao, Audible's chief product and analytics officer ^[16]. Interactive Stories, the Renfield chat, is the one extra the reports tie to generative AI ^[9].

Dracula, 1984 and Exoplanet are all Audible Originals, a label that means Audible's own studios produced the recording ^[14][19][8]. Both chat titles are among them ^[2]. Audible is making this move while facing new competition from Spotify, according to TechCrunch ^[1], in a US market where audiobook sales reached $2.43 billion in 2025 ^[15].

Rachel Ghiazza, Audible's chief content officer, gave the measure in the launch statement. "We're excited to build on these while working hand in hand with creators, because we know that more time spent with stories benefits everyone who helps bring them to life," she said ^[17]. Time spent is the easiest number for a chat feature to raise. A listener who argues with Renfield for twenty minutes adds twenty minutes of time spent and can still abandon Dracula halfway. Neither report includes usage figures, pricing, or whether the extras come with a membership.

The sort I'd apply, to this launch or to any AI extra added to a product people already pay for, has two axes. One is whether the feature fixes a problem users already have in the core activity or adds a new activity. The other is whether it sits inside the core flow or beside it. The speaker label fixes an existing problem inside the flow. Visual Explorer sits inside the flow too ^[18]. Interactive Stories is a new activity beside the flow, a bonus separate from the recording ^[11]. For the in-flow fixes, the useful measures are completion and how often listeners rewind to work out who spoke. For the bonus, the measure I'd want is the share of people who come back for a second conversation and then finish the book.

I'd put the speaker label on every multi-voice title before adding another chat character, and hold the chat at a few books until that second-visit figure exists. The tradeoff is evidence: a test run only on Audible's own productions will say little about how the chat holds up on books from outside publishers ^[2][14].