The old school cassette recorder is funny, right? I had one just like that. When The Beatles album, Let It Be, came out I recorded it on a cassette so I could take it with me everywhere. At night I’d listen to it in bed using the one little ear bud that was the norm for such things back then. Ah, today there are many choices.
I have been trained and have done voice over work (games & training videos). I wanted to do some audiobook recordings for LibriVox some years ago but dealing with the technical aspects of Audacity were tedious and I gave up. So when my cousin begged me (she said Please) to create an audiobook of my SAAI novel the first thoughts I had were how tedious that would be because I assumed I’d do it myself using Audacity. Me. The AI immersed silly woman did not, at first, think of using AI to convert my book. Me. The author who always uses the “Read Aloud” function in Microsoft Word as part of my editing process. I crack myself up.
There are so many options out there for creating audiobooks it’s kind of hard to decide so here’s what I’ve learned in my pursuit of knowledge in that area.
Of course, I eventually get around to asking my AI Assistant for help and was advised as follows.
Q: What are the most secure (not risking my copyright) methods of using AI to create audiobooks?
A. To create audiobooks using AI while protecting your copyright and creative control, you’ll want to focus on tools and workflows that respect authorship, offer licensing clarity, and minimize the risk of voice misuse or unauthorized distribution. Here’s a breakdown of the most secure methods:
🔐 1. Use AI Tools That Offer Clear Licensing and Ownership
Choose platforms that explicitly state:
- You retain full rights to the final audio output.
- The AI voice is licensed for commercial use.
- No claim is made over your content or its derivative works.
Examples:
- Descript Overdub (you own your voice model)
- Respeecher (offers voice cloning with consent and licensing)
- ElevenLabs (check terms—some tiers offer commercial rights)
🧠 2. Avoid Training AI on Your Voice Without Explicit Contracts
If you’re using your own voice or a narrator’s voice:
- Ensure the platform does not store or reuse your voice data for other users.
- Use services that offer voice cloning with consent, and ideally, non-transferable models.
🧭 3. Embed Digital Watermarks or Audio Fingerprints
Some advanced tools allow you to:
- Embed invisible audio watermarks that trace unauthorized copies.
- Use voice recognition tech to detect misuse or cloning attempts.
This is especially useful if you’re distributing widely or concerned about piracy.
📜 4. Register Your Audiobook with Copyright Authorities
Even if AI-generated:
- You can register the final audio file as a derivative work of your original text.
- Include metadata that documents your authorship and production method.
In the U.S., this means registering with the Library of Congress via copyright.gov.
🛡️ 5. Distribute Through Platforms That Offer DRM and Rights Management
Use audiobook distributors that:
- Support Digital Rights Management (DRM).
- Allow you to control pricing, territory, and usage.
- Monitor for unauthorized uploads or piracy.
Examples:
- Findaway Voices
- ACX (Audible)
- Speechki (AI narration with rights protection)
🧾 Bonus: Keep a Clear Paper Trail
Document:
- The AI tool used and its licensing terms.
- Your original manuscript and any edits.
- The final audio file creation date and metadata.
This helps in case of disputes or takedown requests.
Most of that I’d thought about particularly not using my own voice so it could be reused / stored.
One thing not addressed but I wondered about is the location where the application is hosted / licensed / owned.
I came across what seems to be a truly FREE option: Perchange.org But as much as I’d like to save a buck I knew a totally free option is not safe. And checking WHOIS I found they are located in Indonesia, as well. Hmmm… pass.
So next I go looking for unbiased reviews. The first one I found https://publishdrive.com/ai-audiobook-generators-reviewed.html doesn’t even address any of the security interests I have. Go figure.
Not only are there security issues for me but what about the rights of the original voice used by the AI generator? Dive deeper into that with this article: Copyright Laws for AI-Generated Audiobooks: A Global Guide from March 29, 2025.
A couple of decisions I must make before starting are: 1) Will this be for private or commercial use. Maybe I just make a version for my cousin or anyone who asks. And 2) How much do I want to invest in this effort? And not just money but also my time.
From that review article above I learned that Apple offers free audio versions. But – unlike the article leads one to believe and like every other high end site offering this service – they are not truly free. They are Free as in: Come try me for free. KDP also offers audiobook creation through its Virtual Voice / ACX service – from what I read it uses “older voices” as in degraded not aged. Barnes & Noble Press utilizes Author’s Republic which appears to be a self-service narrator marketplace. You have to sign-up just to get all the details. Then, the problem with these is that you don’t have a version you keep – it’s their version for selling on your behalf. Please comment if you’ve experienced something different in this regard.
Back in the day when I commuted more than an hour each way to IBM in Southbury, CT, I was a big time book on Tape / CD consumer. If you have listened to lots of books you know a reader can make all the difference. Given that, the AI voice must be as realistic sounding as possible. So this is definitely a decision variable.
Before selecting a service use whois and scamadviser.com to check your potential service’s website. The latter will give results like this:
Here’s how the record for elevenlabs compares.
ElevenLabs is also located in NYC with offices in various locations like San Francisco, London, and Tokyo.
ElevenLabs is looking like my front runner but then of course there is this:
ElevenLabs was criticized after users were able to abuse its software to generate controversial statements in the vocal style of celebrities, public officials, and other famous individuals,[32][33][34][35][36] particularly attracting attention after users on 4chan used the tool to share hateful messages.[37][15] The software’s ability to closely replicate real voices has raised ethical concerns over impersonation and plagiarism, with critics likening it to deepfaking.[38]
I would dare to write that having had that scrutiny on them only improved and tightened their Ts & Cs. And indeed, there is this. (d) causes me some concerns.
So they can take my novel and create derivative works from, distribute, etc? Hmmmm Not that they would, I’m no Stephen King, but …. Yeah. No.
I like WellSaid. Check out this:
Sounds good, anyway.
But WellSaid is really pricey. $50/mo/user billed annually. Way more than I planned. I tried a sample using one of the ghost stories I recently submitted that has lots of slang and it didn’t sound very good. But, WellSaid appears to give lots of options for fine tuning scripts. There again, how much of my money and time do I want to invest?
I liked Fish Audio but even their heftyist plan doesn’t cover what I need.
Maybe I’m interpreting that information all wrong but I know that 30,000 characters is not that much. A 60,000 word book has got to be closer to 300,000 characters, no?
I can’t decide. I’ve been at this all day. Who knows. Maybe I will do it myself but should I use my own voice when it can be cloned? You can change voices in Audacity – is it enough?
For now, no audiobook.