AI Narration

At the end of April 2022, Google Play Books announced a new beta service for self-published authors. They launched auto-narrated audiobooks that use AI (artificial intelligence) to help authors convert their ebooks into audiobooks. To qualify, the ebooks must be available on Google Play’s Bookstore. Initially, it applied to authors in the USA, UK, Canada, Spain, Australia, and New Zealand, with authors receiving a 52% royalty split. Google has since expanded this to more countries.

This service is currently free to authors while the service is still in beta. This means that Google is still testing and fine-tuning the service. To help with this, they need more content for their AI language models to digest, which is why they’re keen for self-published authors to take up this offer.

Traditionally produced audiobooks, ones using human narrators, are expensive to produce. Most of us can’t afford to hire celebrities like Stephen Fry to narrate our writing. However, as anyone who has tried reading aloud their own work without stumbling, stalling, or sneezing knows, narration is a professional skill. With an average narrator fee of £175 per recorded hour, and a typical 80,000-word manuscript transforming into a nine-hour audiobook, it’s easy to see how quickly costs mount. And that’s before factoring in the fees for hiring a recording studio and any editing.

Yet authors who ignore the audio market are missing out. In 2023, the UK audiobook market grew by over eight per cent and was worth over £222.5 million. Back in 2021, the National Literary Trust surveyed 3,000 UK adults and discovered that two in every five adults listened to audiobooks in their spare time, with nearly a third saying they listened at least once a week.

Accessible Audio

Another reason for creating AI-narrated audiobooks is accessibility. Audiobooks make it easier for readers with sight issues to consume our content. And who doesn’t want more readers?

It’s because auto-narrated audiobooks can be free to produce that allows authors to offer them at competitive prices, perhaps putting them within the reach of those who can’t afford human-narrated audiobooks.

For example, on Amazon, JK Rowling’s Harry Potter and the Goblet of Fire, which is narrated by Stephen Fry, retails at £32.99 in audio format. That’s for just under twenty-one hours of content. The paperback version is currently four pounds.

I tried out Google’s auto-narration with my book, The Positively Productive Writer. The theory is simple. You simply import your text from the ebook file and create the audiobook at the click of a button. It’s best to listen to the text first, which you can do on a chapter-by-chapter basis. Be prepared to make some changes because writing for the spoken word is a completely different skill to writing for the printed word.

Spelling Sounds

Computers have been speaking text for decades, albeit with a robotic accent. However, by 2022, AI narration had a more human-like feel to it. It wasn’t perfect, by any means, but it was certainly acceptable for some uses. Non-fiction books, in particular, can work well with a digitally-narrated voice, where the reader is more interested in learning the information the book has to share.

Google offers over fifty different voices from which to choose, including Archie, a British man aged over sixty, to Padma, an Indian woman aged between thirty-one and forty-five. It’s worth listening to all the voices available and thinking about which you want because there are nuances to be aware of.

Rachelle Ayala (https://www.rachelleayala.net/) is a USA Today bestselling author of playful and passionate romances with a twist. She has used Google’s auto-narration to produce eleven audiobooks.

“It is easy to use and set up,” says Rachelle. “The user interface is intuitive and the instructions are clear. Google’s product allows you to mark off sections and use a different voice. It also allows you to mark words and change their pronunciation precisely by the use of either a recorded clip from the microphone or the characters from the international phonetic alphabet.”

Auto-narration can sound more robotic when it doesn’t pronounce words properly. For example, when it comes across the word read, sometimes it may pronounce it as red, and other times it may express it as reed. If the auto-narration doesn’t understand the context, it may use the wrong pronunciation.

As Rachelle explains, though, it is easy to sort this. Authors can use their built-in microphone and record how to pronounce words properly. Alternatively, you can simply change the text and spell the word phonetically. I found this method worked particularly well as I adapted The Positively Productive Writer. Any changes made to the text in the audiobook version do not affect the ebook text version.

While you can do this on a word-by-word basis, it’s also possible to save the pronunciation across the entire text of the book. For example, if you have a character named Louis, a UK narrator will pronounce it as ‘Loo-ey’ whereas a US narrator will pronounce it as ‘Lewis’. Setting this pronunciation throughout the complete book only needs to be done once.

Rachelle points out that selecting the right narrator is not just about choosing the right voice, though.

“Not all narrators can do all the sounds. For example, the name Michal can only be pronounced by an Indian narrator, Parvati.”

Google’s auto-narration currently offers six languages with additional variations, such as American, Australian, British, and Indian English, and Spanish, for markets in Spain and Mexico.

Nuance Annoyance

A human narrator excels with nuance. The way they say a word can completely change its tone or even that of the sentence. At the time of creating the audiobook of The Positively Productive Writer, it wasn’t possible to emphasise words, or insert pauses.

However, improvements are always being made, and now we can insert pauses of between half a second and two seconds. It’s also possible to adjust the spoken speed a narrator uses for one word or a multi-word phrase.

The auto-narration impressed me. It picked up when dialogue was being spoken and then spoke the phrase “he said” slightly differently, to separate it from the dialogue.

Rachelle agrees that Google’s voices were impressive. “Even way back in 2022, the Google voices were natural and easy to listen to.”

Investing Time

While it may be free to produce the auto-narrated audiobook, there is a time cost. The Positively Productive Writer is over seventy-three thousand words and became an eight-hour audiobook. However, I spent over a month working a couple of hours every day, to edit the text online, ensuring the auto-narration correctly pronounced all the text properly. Ideally, you need to listen to the whole recording again to make sure you have missed nothing!

Rachelle agrees that even though Google’s auto-narration is free, it’s still a lengthy process. “It’s very easy, although time-consuming, to listen to each word. I would love to use it more, but would have to hire editors or spend more time on it, as well as pay attention to new products and platforms.”

Apple Audio

Recently, Apple Books has announced that authors can use their digital speech synthesis technology to create high-quality audiobooks direct from ebook files available on the Apple Book store.

Currently, Apple is limiting their service to limited genres, such as romance, mystery and thriller, and science fiction and fantasy. This will be expanded to the non-fiction and self-help genre soon. All books must be in English.

Authors who use distribution services like Draft2Digital to supply their ebooks to Apple Books can also use this free service, but must access it through the Draft2Digital platform.

Audio Acceptability

Once created, Google Play Books allows authors to download their audio files and distribute them elsewhere, on condition the audiobook is available in the same country on Google Play Books. While that sounds exciting, not all platforms accept auto-narrated audiobooks.

However, in December 2023, Findaway Voices, a distribution service for audiobooks, announced it will now accept audiobooks created using Google Play Books’ auto-narration service. As with all things AI-related, the market is constantly changing and adapting, something that is bound to increase further now that Findaway Voices is owned by streaming service Spotify.

Audiobooks that use auto-narration will not replace human narration completely. Instead, think of it as the same text in another format, giving readers more choice. It’s no different to our text being offered in paperback or hardback formats. Some readers prefer the quality of hardbacks, others prefer the price point of paperbacks.

Auto-narration has not only allowed me to produce an audiobook version of The Positively Productive Writer, but I’ve priced it at under five pounds, which is just a few pounds more than the ebook version, yet a few pounds less than the paperback. Different price points suit different pockets.

Rachelle’s audiobooks range in price depending upon length. A short novella of under an hour is free, whereas her nine-hour audio version of Summer Love Puppy is less than three pounds.

Audio is changing, and auto-narrated audio is still developing. Looking forward, Rachelle has her hopes. “I would love it if the AI could read it, give an error report and point out questionable passages, where the author can then pinpoint and make adjustments or okay it.”

Chances are, it won’t be long before Rachelle’s hopes come true. In the meantime, when it comes to the business of writing, moving into the audio market is becoming more accessible, particularly for self-published authors.