And why we think it matters
If you have browsed audiobooks recently on Apple Books or Amazon, you may have noticed something odd. Some titles now carry a small label: “narrated by a digital voice” or “AI-generated narration.” The number of these titles is growing fast. Apple began offering AI-narrated audiobooks in 2023, and Audible has since partnered with publishers to convert print books into AI-narrated audio at scale, with over 100 synthetic voices available in multiple languages.
Spotify has gone further still, launching an AI audiobook creation tool (powered by ElevenLabs) that actively encourages authors to generate synthetic narrations and publish them directly to the platform. The direction of travel is clear: more AI, faster, cheaper.
The argument for AI narration is straightforward: it is cheap and fast. A professional human narrator costs thousands of pounds to record a full audiobook. AI can do it for under £50. That means more books can become audiobooks, including titles that would never have justified the cost of a human recording.
We understand the logic. We just aren’t convinced it makes a good enough audiobook – and it shows a lack of support from creatives that we have come to expect from these big tech brands.
What you actually hear
AI narration has improved enormously in the last few years. The best synthetic voices sound smooth, clear, and technically competent. For short passages, you might not immediately notice the difference. But audiobooks are not short passages. They are 8, 10, 15 hours of sustained listening, and over that length the differences between human and AI narration become impossible to ignore.
A human narrator brings interpretation to every sentence. They know when to slow down before a reveal, when to let a pause do the work, when a line of dialogue needs warmth or menace or wry understatement. They make choices about pacing, tone, and emphasis that reflect a deep understanding of the text. The best narrators do not just read a book: they perform it, and that performance is a genuine part of the experience.
AI does not interpret. It processes. It can mimic the surface features of human speech (intonation, rhythm, pitch variation) but it does not understand what it is reading. It does not know that a sentence is funny, or devastating, or that the narrator is lying. It cannot land a joke. It cannot convey the quiet tension of a character who is about to make a terrible decision. It just reads the next word. And the ability it does have to mimic those surface features was built by training on copyrighted recordings and the past performances of real narrators, often without their knowledge or consent, and with no compensation to the rights holders whose work made the technology possible in the first place.
Where it matters most
For some types of content, the gap is smaller. A straightforward non-fiction book with short chapters and simple sentence structures can sound passable with AI narration. But even here, “passable” is the ceiling. You notice it in the transitions, in the way the voice handles lists, in the absence of any real warmth or personality.
For fiction, the gap is enormous. Characters need distinct voices. Dialogue needs timing. Emotional scenes need a narrator who can make you feel something, not just hear something. A flat, synthetic reading of a powerful scene does not just miss the mark: it actively undermines the book.
Memoirs are perhaps the starkest example. When an author reads their own story in their own voice, the audiobook becomes something print cannot replicate. You hear the hesitation in a difficult memory, the laughter in a funny one, the specific way someone says a name that meant something to them. AI is not a viable susbstitute.
There are better ways to innovate
We are not against technology. We are a technology company. Our x-book format uses sophisticated indexing to synchronise human-narrated audio with the corresponding text, letting listeners see the words on screen as they listen, look up vocabulary, search within the book, take notes, and share quotes. That took serious engineering and it genuinely improves how people experience audiobooks.
But it improves the experience without removing humans from it. The narrator is still a real person. The author’s words are still performed by a skilled professional. The technology adds a layer on top of a human performance rather than replacing it.
That is the difference. Big tech platforms are using AI to cut costs by cutting people out of the process. We think the more interesting question is how technology can make a human-narrated audiobook even better. The x-book is our answer to that question, and we think it is a more honest kind of innovation.
Our position
Every audiobook you will find in the xigxag catalogue is narrated by a professional human voice. That is the standard we hold ourselves to, and we actively work to maintain it.
This is a deliberate choice. Right now, the quality is not there, the compensation model for the narrators and rights holders whose work trained these systems is not there, and we do not think listeners are well served by platforms that prioritise scale over experience. If those things change, we will look again. But today, human narration is a fundamental part of what makes an audiobook worth listening to, and we are not willing to compromise on it.
We know this means our catalogue will never be as large as platforms that use AI to convert every available title into audio. We are comfortable with that trade-off. We would rather carry 120,000 titles that all sound excellent vs. 500,000 where half of them sound like robots.
How we enforce it
Being honest, it takes more work these days than it should. We do have one narrow exception on the platform, a set of IT governance instructional manuals distributed on behalf of a specialist education publisher. These are AI-narrated x-books that are not merchandised in our catalogue and are only accessible through the publisher’s own website. Our regular users will only encounter them by searching for them specifically.
Beyond that, our position is zero tolerance, and we actively enforce it. Publishers sometimes submit AI-narrated content to us. We reject it when it is properly labelled. The harder problem is when it is not labelled. AI narration is getting good enough that it can slip past initial checks, and not every publisher is upfront about what they are sending. When we identify AI narration that has made it onto the platform without being flagged, we pull it from sale.
We also ask our customers to help. If you are listening to something on xigxag and you think the narration sounds synthetic, let us know at support@xigxag.co.uk. We check every report and we take it seriously. We would rather pull a title and investigate than leave something on the platform that does not meet the standard we have set.
How to tell the difference
If you are buying audiobooks elsewhere, here are some things to look (and listen) out for:
Listen to the sample. Most platforms let you preview the narration before buying. Listen for how the voice handles a transition between narration and dialogue, or a shift in emotional tone. AI tends to sound smooth but flat across these moments, while a human narrator will reflect the shift in tone.
Look at the publisher. Major publishers with established audio programmes (Penguin Random House, HarperCollins, Hachette, Pan Macmillan, Bonnier) still overwhelmingly use human narrators for their front list titles, and they are taking a leading role on appropriately identifying AI narration. AI narration is more common on platforms with open or self-service upload models, where there is less oversight over narration quality.
The bottom line
Humans have been telling stories with their voices for tens of thousands of years, long before anyone wrote anything down. The written word is only about 5,000 years old. The human voice came first, and it still does something that no technology has managed to replace.
AI narration exists because it is cheap, not because it is better. For listeners who care about the quality of the experience, human narration is not a luxury: it is the point. A great narrator can make a good book unforgettable. AI cannot do that yet, and we are not convinced it ever will.
Every audiobook on xigxag is narrated by a real person. Browse the catalogue at xigxag.co.uk and hear the difference for yourself.
