My Voice or AI Nother’s? A tale of three choices

The Audio Version
The first choice was only partly mine. Publisher Adlard Coles was planning an audio version of Stars to Steer By. Who would read it? They suggested three professional narrators which I put to the family vote. One, whose voice was rather light (and pleasant, I thought) was rejected by all except for myself and my oldest granddaughter. The other two -- stronger, more assured, more (or less?) flexible -- split the voters entirely: Francis, who listens to a lot of audio books and who spent a generous amount of time researching samples beyond  the clips we had been sent, headed one faction but the Adlard Coles editorial office AND an equal number of family voters, preferred the other.

I began to feel both protective and shy. Stars to Steer By is non-fiction, it’s historical, it’s researched but it’s also, occasionally, personal. My thoughts, my history, my words. I discovered that I wanted to speak them, some of them anyway… the opening for instance: ‘One early evening in August, I looked across St Katherine’s Dock at the yacht Maiden, lying calm and elegant in the gentle sunlight.’  That had been an emotional encounter as well as a significant moment in development of my thinking. I wanted to read it.  

Uncommon Courage, my previous book for Adlard Coles, had clearly needed a male narrator (WW2 war at sea) but I’d been allowed to top-and-tail it in my own voice. Subsequently my confidence had been developed in the harder school of writing and narrating a short film, We Fought Them in Gunboats, for director Tim Curtis. I remember one particularly bad day when we were about to hit a deadline and I just couldn’t get the words and inflexions right. Tim got fed up and sent me out for a walk. I snuck down to the Woodbridge cinema bar and had a bowl of cheesy chips and the best glass of red wine ever. Back to the studio and the job was done in a twinkling.

This undertaking would be so much bigger – 120,000 words, 4 days recording, 30,000 words a day. I don’t have a strong voice, would it hold out? I’m not a professional, would I bore the listeners with insufficient variety of tone? And worst of all, would I slip into what the children call my ‘telephone voice’? (Think old-fashioned BBC Third Programme but not genuine.) I told Francis and Bertie, but not the wider family, that I was thinking of putting myself forward, I couldn’t face their potential embarrassment. I also had a stubborn cold so continued to prevaricate until Kate (lovely desk editor at Adlard Coles) slapped down an ultimatum. She was about to have a few days off and the audio department needed to decide who would narrate the book. I blew my nose, sat down on the bed, read 4 ½ minutes of the introduction into my phone and sent it over.

Lucy Wroe
woman of many talents
https://www.lucywroe.com/
A couple of weeks later, Francis delivered me to Wardour Studios, my bag clanking with cough medicines, throat pastilles and nasal sprays. There I met Lucy – smiley and immediately reassuring. I’d been promised an experienced producer and that promise had been kept. Lucy’s a musician, a children’s story reader, a podcast producer and broadcaster. Her voice is light, clear and flexible (she’d have made a lovely job of reading my book). I knew I could trust her to help me with the second choice: how best to read other people’s words?

Stars to Steer By is the story of many people’s stories: from Anna Brassey circumnavigating the world with her family and staff in 1876/7 to Vuyisile Jaca, sailing as part of the Maiden team in 2023/4. I quote other women as often as I can. That’s clear enough on the printed page but how should I make this audible? One way was to emphasise a characteristic; I annotated my script ‘assertive’, ‘energetic’, ‘angry’, ‘youthful’. I googled accents and begged Lucy to stop me if they sounded naff. She promised me that she would. ‘If in doubt, always read in your own voice,’ she said.

I searched the internet for recordings of the women I had quoted. It wasn’t an unmixed blessing. I was excited to hear Pamela Bourne Eriksen interviewed on the wreck of Herzogin Cecilie https://www.youtube.com/watch?v=KzoNPPseT3U but if I’d tried to reproduce her high, upper-class 1930s diction it would have sounded like caricature and would have obscured, not clarified, her passionate, determined personality. Or so it seemed to me. 

Lucy and I sat in adjacent boxes talking through headphones. There was no outside world, just my voice and the words.  I loved the concentration, winced at clumsy sentences and was shocked by the number of errors that had escaped the repeated scrutinies of the copy editor, proof reader, Kate and me. Lucy told me this was normal. She said she can’t understand why the audio narration isn’t done immediately before the text goes to the printer as it’s such a good way of hoovering up the final typos. On my final day in Wardour Studios I met author and podcaster Lucy Meggeson recording her forthcoming book Thrive Solo, a celebration of being single and child-free.  She too was loving reading her own words, was shocked by her previously-unspotted typos - particularly when she discovered that she’d written ‘barrister’ for ‘barista’ describing herself during an impecunious period in her life!  

Lucy Meggeson
podcaster
reading Thrive Solo
To finish, I needed to read a prohibition against the recording being used to train AI bots. This reminded me of a current dilemma, not connected with the luxurious experience of reading Stars to Steer By. Adlard Coles is part of Bloomsbury, a company with the resources and infrastructure to support audio recordings of even books like mine, unlikely to become runaway best sellers (but if you’d like to order, here’s the Spotify link  https://open.spotify.com/show/6iv6mwi3X9m8G5ZxHTkotn?si=cdfedf26bdc841e2 or visit Audible). Golden Duck, my home publishing company, isn’t in that position. Even if reading might be done for love (as blessed Anna Bentinck did for The Salt Stained Book, volume 1 of the ‘Strong Winds’ series) the up-front studio and production costs are beyond us. Yet I really want people with reading difficulties and partially-sighted people to be able to access our books. Pebble (Strong Winds series v6) is a particular yearning as it tells the story of a child losing his sight. What will there be for him to read?

This meant that I didn’t immediately press DELETE when I recieved  message from Amazon KDP

Congratulations! You’re invited to participate in KDP's beta for audiobooks. Starting today, you can now produce audiobook versions of your eligible eBooks using virtual voice narration and reach new customers by making them available on Amazon, Audible, and Alexa. Customers have already enjoyed listening to millions of hours of audiobooks with virtual voice from KDP authors.

Bertie investigated, using A Ravelled Flag (volume2) as sample. The voices on offer were not too bad at all, he reported back. But did we want to go even further into the power of Amazon, using robot voices to read stories that are even more personal and close-to-the heart than my non-fiction books? Lucy described one way that these AI readers are being trained. Studios are hired, actors recruited, paid LARGE fees, then set to read strings of words – Wikipedia entries, for instance – anything not protected by copyright.  Despite the fees, many actors are refusing, knowing that that they are not just doing a day’s work for a day’s pay, but selling their voices in perpetuity. They will never know what texts their cloned voices will be used to read.

The Strong Winds series

If a bot read Pebble or A Ravelled Flag more people would potentially be able to enter the world of those stories – which, after all, is not a ‘real’ world. I hope that some people already read the books to each other. I have no control over that. So why is it that I find this third choice so difficult? How has it become an ethical, not a pragmatic issue? All the books are available on Kindle which means basic text-to-speech will already be enabled. Why should people who cannot read print, be expected to remain satisfied with that mechanical device?

I don’t have an answer to that second question but am clearer now about the first. What Lucy and I were doing, in our separate studio boxes, was a creative reinterpretation of Stars to Steer By, if that doesn’t sound too pompous. A 'live' reading is a little like playing a piece of printed music, it's going to be slightly different every time. 

At my grand daughter’s christening last weekend I met a couple who only EVER listen to audio books if they are read by the author. ‘We want to trust that they really know what they are saying. With the other narrators, they’re just reading words, aren’t they?’ I was a little startled by this radically purist approach but essentially I had to admit that was why I’d wanted to read Stars for myself and had sent in my initial rather scruffy phone sample. But that could have gone no further without the additional input from Lucy and her colleagues -- which I can't afford for my Golden Duck titles.

I'll have to say no to the KDP offer and wish that all the ingenuity and money that’s being put into cloning voices could instead be poured into human narrators and producers. Then people like my child hero Liam, in Pebble, (or myself and my contemporaries as our aging eyesights fail) wouldn’t be condemned either to exclusion from the printed world, or access by bot.


Thank you Adlard Coles & Wardour Studios for an unforgettable experience.

Stars to Steer By was published on May 8th 2025. Also available via  Audible and Spotify. 

 

Comments

This is so interesting, Julia! I am an audiobook virgin, neither an avid listener (I prefer reading printed copies), nor having had any of my titles produced in audio while they were under contract to their original publishers. So I now have a dilemma, too... use a virtual voice to read the books, or continue to have them all unavailable as audio. Given that most of mine are for younger readers, audio seems a good thing - but studio production for my backlist titles is beyond my budget so I am experimenting with AI. The KDP voices allow editing so that any mispronunciations (fantasy names are a minefield!) can be corrected, and speed can be adjusted on a word for word basis, as well as by the listener after publication I believe? Yes, there's a principle here. But AI has already trained itself on most of my books, so maybe it's time to make use of the tech rather than resist progress? I still hope that every creator - writers and voices/actors included - will eventually be paid something for the use of their copyrighted works to train AI, and then maybe we wouldn't all be made to feel so guilty using it?
Julia jones said…
Thank you Katherine. I was so close to saying yes, both for the budget reasons and for the accessibility 9time to make use of the tech rather than resist progress' as you say. And yes I want my stories to be heard. I would love to read them myself, as I would to my children at bed time but can now see, more clearly than ever, how much difference is made by the studio and producer input. And if one is then asking people to pay, the quality must be there - otherwise it's like sending out a book full of broken lines and punctuation errors. I need a way forward and at the moment, am not seeing it. Look forward to hearing more about your experience.
Forgot to mention one positive about virtual voice is that the resulting audiobook can be published quite cheaply, hopefully making it more affordable for listeners who want something better than text-to-speech but haven't a budget for the whole studio experience. It's not really like sending out a book full of errors because it works from the original text, so provided the text has been well proofed then the audio seems a pretty good reflection - though if the text does contain errors, then these will also be faithfully reproduced in audio! Hence the editing facility, which requires human input. Apple have a similar offering, where they do the checking/editing before publication.