I've been talking about AI narration for several years now, but it's just starting to go mainstream and I've been getting emails every day recently asking the same questions, so this is a round-up article with the most important information.
For context, I am an audiobook narrator. I narrate my own non-fiction and short stories. I absolutely value human narrators, and I have spent tens of thousands of dollars hiring professional narrators for my novels and non-fiction over the last decade. I am also a futurist and I embrace AI tools as part of my creative and business practice.
This episode covers:
- Why are AI-narrated audiobooks an important development?
- “AI voices are robotic and don't have the right intonation. No one wants to listen to that.”
- Will AI-narrated audiobooks put real human narrators out of a job?
- What services can you use to create AI-narrated audiobooks? How much does it cost?
- Can anyone create an AI-narrated audiobook? What intellectual property rights do you need to hold?
- Where can you sell AI-narrated audiobooks? Can you just upload them everywhere?
- What price should an AI narrated audiobook be?
- How can you make sure that listeners know this is AI-narrated? Why is labeling important?
- How do you market AI-narrated audiobooks?
- Where is AI narration heading in the next few years?
This episode is sponsored by my patrons at Patreon.com/thecreativepenn. Thank you for enabling me to continue exploring the future of creativity and the author business model.
I'd love to know what you think. Please leave a comment here or tweet me @thecreativepenn or email me if it's more personal.
Why are AI-narrated audiobooks an important development?
Most content in written words is not available in audio format, primarily because it is expensive to produce, and many countries and languages don't have an established audiobook production ecosystem.
It's important to have written word content in audio for accessibility reasons, but also because many people choose to consume in audio. It's easier to listen than read with your eyes because you can do other things at the same time, driving, exercising, chores, etc.
There is room for all formats and our job is to make sure our books are available in whatever format the customer chooses. Personally, I listen to a lot of non-fiction audiobooks while exercising, cooking, cleaning, and doing chores, as well as when traveling to rest my eyes. My husband listens to epic fantasy audiobooks every evening for relaxation.
Audio content also needs to be more diverse. Why do I always have to listen to business books read by a US male voice? Why can't I choose the voice of the narrator?
People identify with people who speak like them. Why can't someone choose the accent they prefer as well as the language and gender in the same way we choose the voice of our Google Assistant, Siri, or Alexa?
It would be prohibitively expensive to create multiple versions of audiobooks with different voices, but it could be achieved with AI narration.
You can listen to Your Author Business Plan narrated by me (British English female).
You can also listen to the same audiobook read by Mia from Google Play, an American female voice.
“AI voices are robotic and don't have the right intonation or emotion. No one wants to listen to that.”
AI narration will never supplant high-quality audiobooks performed by actors and voice artists at the pinnacle of their craft.
They will never replace the ‘art' end of the audiobook market.
They are not intended to.
In fact, the rise of AI narration might even push human narration into even more beautiful forms of audio art, because the ‘mass market content' side can be produced with AI, and narrators can focus on the books they care about.
Remember, you are not your audience, and you don't speak for the entire world of potential listeners. Even if you never want to listen to an AI-narrated book, why stop others from listening? Many people would rather have the content in audio in any voice than not at all, especially if their language is under-served in terms of audiobooks.
AI voices are also improving all the time. I've included samples here so you can listen to some of mine.
I have two versions of A Thousand Fiendish Angels, a short story trilogy. Here's me reading as a British English female with an accent from the South West of England.
The following version is digitally narrated with the voice of William Birch, a British male also with a southern English accent, produced by DeepZen Limited. I like the narration, and it's more interesting because it's a male voice. Two of the three stories have male protagonists and I love how different the stories sound from mine.
You can buy A Thousand Fiendish Angels here direct from me and listen on the Bookfunnel app. It will also be available on multiple retailers in coming months as it's distributed through DeepZen.
Human narration is also variable, as is listener preference.
I'm also an audio consumer. I listen to a LOT of audiobooks, almost all non-fiction, and I often wonder if I am listening to an AI, anyway. There are plenty of human narrators who sound expressionless, especially for the more technical or business books I listen to. Plus, I listen at 1.5x speed anyway, so it often doesn't even matter.
There are also an increasing number of tools for other voice-first needs, of which audiobooks are perhaps the smallest use case. Gaming has embraced AI tools for years, including AI for voice.
Podcasters have been using AI tools for several years now. I created a Voice Double using OverDub through Descript.com, which can be used in the podcast editing process. Mark Leslie Lefebvre and I even had a conversation between our Voice Doubles here. We need to redo it as the technology has moved on in the last 2 years and the voice doubles are so much better now.
Will AI-narrated audiobooks put real human narrators out of a job?
AI narration will never replace high-quality human performance, or the nuances of an actor's ability to communicate meaning, or narration by the human author.
But the time is right for a stratification of audio rights into several streams — the full cast multi-voice audio production, the human single or multi-voice read, and the AI narrated version, of which there might be several versions.
In this way, each audiobook is no longer one product, but many, in the same way that we have multiple versions for ebook, paperback, hardback, large print, and other formats.
It may change the job for some narrators, but in the same way that the job has changed for many translators with the advent of AI tools like Deepl.
Many translators have more work now, as there are more books being translated because of AI. The bulk of the work is done by the AI engine and then the translator brings the art and skill of finessing the translation into a finished product.
As voice licensing becomes more common, narrators will license their voices [as Samuel L Jackson did for Alexa), attracting other revenue streams without the workload of narration.
They might use this technology to create the ‘first draft' of an audiobook and then finesse the finished audiobook in the same way that AI-augmented translators do. Or authors and creators will be able to license their voices for creating audiobooks themselves.
Narrators, check your contracts. Who owns your master recordings and what can they do with them? Could whoever hired you create an AI voice from the recordings and license that?
Estate management will also expand to include voice licensing for talent after death.
But what do narrators think?
You can listen to narrators and voice artists discuss AI narration on the VOBoss Podcast. They have lots of different episodes on the topic, and there are a variety of views from ‘it should not be allowed,' to ‘it all sounds terrible and AI can never communicate like a human,' to ‘we should embrace AI and license our voices and use the tools to spread our personal brand and make more money.' As ever, there are always varied opinions on new tools and technologies.
Also, check out VoiceBot.ai which also has a podcast discussing these issues in depth.
Publishers Weekly (4 Feb 2022) rounds up a variety of responses to AI narration, with concluding comments as follows.
Hillary Huber is a member of the SAG-AFTRA [Screen Actors Guild–American Federation of Television and Radio Artists] Audiobook Steering Committee, and a board member of the Professional Audiobook Narrators Association (PANA.) She says, “AI is coming. We can’t stop it, but hopefully we can be proactive by creating protections and, more importantly, by raising awareness in general so that listeners and authors can make informed choices.”
Anthony Goff (previously senior V-P and publisher of Hachette Audio said, “The technology is good and getting better but can never fully replace the art of audiobook creation… With the amazing growth of audio over the past few years, it’s clear that consumers love narration by professionals, and we do not see AI replacing that—rather, we see it supplementing it.”
Absolutely. I agree with both of these comments. AI supplements the existing audio market, and education is needed for consumers so they know what they are getting, and for narrators so they can use the technology to expand their creative and business potential, while still retaining the high-end human production values we love so much.
There is room for all, especially if we start stratifying audio rights in contracts, and educating listeners about their different options and the impact of each choice.
What services can you use to create AI-narrated audiobooks?
There are different services with different levels of help and distribution options.
Google Play Books is free (as of May 2022). They have an easy-to-use dashboard where you claim your ebook published on their platform, choose the voice, and then review the audio. It generates in a few hours and you can download the files and sell them elsewhere if you like.
This is the Dashboard now, so you can see how easy it is to use. Even if you use a different service e.g. PublishDrive to distribute to Google, you can log into the dashboard and claim your books for AI audio.
As an independent author, your ebooks have to be wide to access the platform.
[You can listen to my discussion with Ryan Dingler from Google Play here, where we discuss AI for audio toward the end. I will have another interview with Google Play Books in the next few months for an update.]
DeepZen is a premium service with ethical standards where they pay narrators for voice licensing. They also have distribution to most audiobook services because their quality is so high (excluding Audible). I've produced two audiobooks through DeepZen, which are in their distribution queue at the moment.
There are lots of other AI narration and production services. Check out AI comes to audiobooks, Publishers Weekly, by Thad McIlroy for a list and opinions on AI voice in the traditional publishing industry.
PLEASE make sure you read the terms of service before you use any platform. Download a version and keep it as evidence of what you signed at the time, as terms will inevitably change.
Can anyone create an AI-narrated audiobook?
If you own and control the audiobook rights for specific intellectual property assets (i.e. books or short stories), then you can use AI narration.
Some of the services are not available to all countries, but they will likely expand over time. Most are starting with English and some of the most common languages, and again, will hopefully expand this.
If you have already licensed your audio rights, you cannot create an AI-narrated version.
Have you signed a contract with a publisher or an audio production company? It's likely you have licensed audiobook rights in their entirety, but check your contract to be sure.
You can't create an AI version if you have signed an exclusive license with ACX or another service. That exclusive license also covers all audiobook rights.
I think we need to start negotiating for the stratification of audio rights in contracts, so consider negotiating specific audio rights, e.g. specify a human narrator, or AI production, depending on the situation.
Where can you sell AI-narrated audiobooks? Can you just upload them everywhere?
If you create your audiobook with Google Play, you can sell it on Google Play immediately.
You can also download it and sell it direct. I currently use Payhip with Bookfunnel integration, but many authors use Shopify, Gumroad, and other services. Audiobook listeners can listen on the Bookfunnel app, which is really easy and Bookfunnel provide customer service for anyone who has issues.
DeepZen have distribution to most audiobook services as their quality is so high, but you have to produce and distribute through them to access this, and it is a premium service.
You cannot publish AI-narrated audiobooks on ACX or FindawayVoices as of mid-May 2022. This may change in the future, but for now, it's human narration only.
What price should an AI narrated audiobook be?
It's up to you.
However, it is still the same content, which is valuable, so don't make it free, unless it is a promotional price or a first in series.
I think it should be cheaper than the human narration, but not free.
Is anyone buying the AI-narrated audiobooks?
It's early days, and most authors are just trying things out so far. I've sold a few but I tend to push my human narrated audiobooks more heavily, especially as I am the narrator of many of them!
Some authors are embracing this technology wholeheartedly. Here's a comment from my YouTube channel from romance author, Alex Anders.
“This month I converted 65 of my fiction stories to AI created audiobooks on Google Play. They have outsold my ebooks in territories they are available. It blew my mind when I realized what was happening.”
In a subsequent comment, Alex said, “I sold 6 sexy romance AI generated audiobooks yesterday and the only market it is available on is the US Google Play store. I sold 0 ebooks in the same territory. I think we are on the verge of a serious revolution and it blows me away.”
Clearly, sales will increase if there is more AI narrated content available, available in more markets, in more languages, and if listeners are educated about their choices.
How can you make sure that listeners know this is AI-narrated? Why is labeling important?
I think AI-produced audiobooks should be labeled. This is part of the ethical use of AI tools. There's no need to hide the fact an audiobook is AI-narrated, so why not make it obvious?
We don't want to mislead customers, and in fact, we want to educate them about their choice. Do you want this audiobook narrated by a human? Or do you want this narrated by an AI voice? There is room for both in the market.
I'm using a ‘sticker' on my audiobook covers with Digitally Narrated on them. I hope something like this becomes the industry standard. We can also use something like it to emphasize human narration and make that part of the packaging as well. I'm using a banner for “Narrated by the author” to differentiate the product.
How do you market AI-narrated audiobooks?
In the same way as you market any other audio or ebook that you sell direct. I have a section on my Payhip store for digitally-narrated audio.
Make sure you link to it on your author website, and in email and social media. Tell your readers about it and educate them on why you're using AI for voice.
Where is AI narration heading in the next few years?
AI narration will become an increasingly accepted part of the audio ecosystem, and services that don't accept AI narrated books now will likely do so in the next few years — unless their point of difference is specifically artistic, human narration.
I think Audible will allow AI narrated content at some point. Given they are owned by Amazon which already has AI for voice with Alexa and Polly, they will likely produce their own software-as-a-service in the same way that Google Play has.
Spotify bought Findaway in November 2021 and Spotify allows AI-voices for music, and AI-collaborative musicians, so why not allow AI-narrated audiobooks?
In the same way that consumers expect an ebook version of every print book, they will also expect an audiobook version of every ebook and written word resource, in every language, and perhaps in varying accents within that language.
We can only achieve this with AI narration because there are not enough humans to narrate in every language and every accent.
There are already apps that read written content aloud in various voices, and these might acquire licenses to read audiobooks as well as other content.
Many countries do not have a developed audiobook production ecosystem, so AI narration will help them produce massive amounts of content in audio. Because of this need and the resistance to AI audio in English, the growth of AI narration will primarily be for markets other than English over the next few years.
This will be fantastic for audio-first consumers as all content will be available in audio and in languages they prefer.
For creators, rights-holders, and narrators, AI-narrated audio will drive down costs — but it will also drive down revenue.
Ubiquitous audio and the rise of subscription models will continue to erode income from audiobooks, some of which may be made up in increased volume, but not all.
This digital abundance only increases the importance of personal branding and developing a relationship with readers and listeners, as well as expanding streams of income.
This is the heart of the emerging Creator Economy.
Don't rely on streaming and subscription services for your audio revenue, in the same way as many of us no longer rely on the big services for the rest of our author income. Sell direct and use your intellectual property assets to create new products and new revenue streams.
Readers and listeners love their favorite narrators and authors, and will continue to buy from them, so emphasize high-quality human narration options as well as AI audio. Charge premium prices for the human version in the same way we charge more for a limited edition hardback compared to an ebook borrow on a subscription service. Audiobooks shouldn't be just one product.
Create multiple streams of income with your audio by selling on different platforms, not just on the subscription and streaming services. Emerging blockchain and NFT opportunities will enable limited edition collectible digital audio for super-fans, and you can use patron models, crowdfunding, and other revenue streams.
Those who embrace AI narration for audio — as well as high-quality human narration — will reach more listeners and expand their market.
As Kevin Kelly says in The Inevitable, “This is not a race against the machines. If we race against them, we lose. This is a race with the machines. You’ll be paid in the future based on how well you work with robots … It is inevitable.”
Let's figure out how we can work with the robots (AI) in order to create more, serve more customers, and increase the value of our creative businesses as authors, narrators, and rights-holders.
Personally, I will continue to narrate my own non-fiction and short stories, and I intend to license my voice when that becomes a possibility. I'll also continue to hire professional human narrators through Findaway Voices, and I will also produce multiple forms of AI-narrated audiobooks. I'm more excited than ever about AI tools for authors, narrators and rights-holders!
I'd love to know what you think. Please leave a comment here or tweet me @thecreativepenn or email me if it's more personal.