Audible’s plans for AI narration, translation and end-to-end production will mean listeners lose out, argues Infotec’s Simon Guerrier
Last week, the Amazon-owned audiobook company Audible announced plans to launch ‘fully integrated, end-to-end AI production technology’ to offer narration and translation by AI, and to automate the entire production process of making an audiobook.
Photo by Jarrod Reed / Unsplash
Such services will be made available through ‘select partnerships’ with certain publishers.
Bob Carrigan, CEO of Audible, said the company aimed, ‘to expand the availability of audiobooks with the vision of offering customers every book in every language, alongside our continued investments in premium original content.
‘We’ll be able to bring more stories to life — helping creators reach new audiences while ensuring listeners worldwide can access extraordinary books that might otherwise never reach their ears.’
But as well as being a tech journalist with an interest in AI, I am a writer and producer with more than 20 years experience in audio drama and documentaries. And I don’t think this move by Audible will be good for listeners – or for creators, either.
I was asked by the Writers’ Guild of Great Britain to explain why not. Here is my response:
‘AI narration and production of audiobooks means a worse product for listeners. Despite improvements in artificial speech, it still sounds odd and jarring, especially at any length, which means we can’t lose ourselves in the story being told.
‘AI narration and production of audiobooks is also bad for the creative industry. A good audiobook requires a skilled reader and production team to bring out the nuance and meaning, creating a direct connection with the listener. That skilled work is worth paying for.
‘And AI narration and production is bad for the people who use it to produce audiobooks: they produce a worse product, don’t develop their own skills and serve only to add to the squeeze affecting the publishing world more generally.’
All this, of course, takes place in the wider context of AI’s impact on the creative industries. Just last night, the House of Lords voted down the government’s Data (Use and Access) Bill for a second time. Cross-bench peer Baroness Kidron, who led the move against the Bill, says that the proposed changes to copyright legislation to facilitate AI effectively ‘redefine theft’. She also called for greater transparency in the way AI companies use copyrighted materials to train systems, with a requirement to seek permission from copyright holders before use.
In related news:
Public sector digital transformation is too much, too soon, too disconnected
High adoption of AI produces little economic gain – new study
As a focus-struggling reader who benefits greatly from audiobooks, I have two issues, largely relating to immersive reading where you have the audiobook and the book at the same time.
1. Having an audio narration available once you have paid for an ebook – it is prohibitively expensive for many to pay twice. There should be a way of having book+audio at the same time without huge cost. So people are turning to A.I. and third party tools like Speechify and ElevenLabs to bypass the cost of paying for the narration.
2. SO MANY BOOKS are not available with any narration at all. It is astonishing why so many books remain audiobook free.
So to me, one of the big questions is how do you dramatically enlarge the availability of audio content WITHOUT using A.I. It’s not perfect, but for me, it’s 95% there with many books, and we have only just begun. It has opened up some heritage or unloved old books that are completely off the publishers/Audible radar.
I have to say, ElevenLabs doesn’t sound odd and jarring to me. I can listen to it for hours… so from an end user’s perspective, it’s actually working out well. Perhaps the inclusion of A.I. voice into Kindle (rather than Audible, and rather than the rubbish electronic voice it currently has) into books will be a game changer in opening up ALL books to ALL readers (and I would think, though I’m not expert, readers with vision disability?)
I appreciate this is probably a hot take and not one that will go down well.
Well, I think point 1) “I don’t want to pay for an audio version” explains point 2) “Why aren’t there more audio versions?” There *is* a cost in using AI – in energy, in water consumption, in the wider environment. It comes back to the same issue: in “opening up ALL books to ALL readers”, who pays?