AI and Audiobook Narration

 

For a long time, audiobooks were something I’d only listen to on a driving trip. On cassette tape. Yep, it was that long ago.

I believe that books should be available in whatever format people want to consume them in: print, audio, ebook. I was very happy when my first three books came out in audio format. But, as I noted above, I didn't listen to audiobooks on a regular basis.

Sometime in the last couple years, I started consuming more non-music audio content: podcasts, audio dramas and books. The books started when I wanted to read all of the Dark Shadows novels originally issued in the late 1960s, early 1970s. I had a few paperbacks that I bought when they first came out, but the majority of them I didn’t have access to until they were reissued as audiobooks. Read by one of the actors from the original soap opera, Kathryn Leigh Scott, they were all available for checkout from my local library on Hoopla Digital. It took me awhile to get through them but, after I finished, I realized I enjoyed them so much I wanted to check out other audiobooks.

I've listened to a variety of books since then with different narrators. The most important thing is the content. If you don’t have a good book, you’re not going to have a good audiobook. I did discover, though, that the right narrator for the project could enhance my enjoyment of the story.

In the past week, I heard about the use of artificial intelligence in audiobook narration. Google is getting in on the act as well as Speechki and DeepZen. Google Play Books has a beta version available and is working on making it available to publishers. You can check out some free audiobooks created using their software here. These are all books in the public domain.

Speechki is a recording platform that claims to be able to produce an audio book using AI synthetic voices from input in 15 minutes. The publisher or author uploads their book and selects a voice. DeepZen has a stable of actual, real voice over artists who have contributed their voices in some way, don’t know the details. From whatever sampling is done, DeepZen takes an input file of a book and does their magic to create the audiobook using a voice selected by the publisher or author of the book.You can go here and listen to some samples. I’ll wait for you.

Ah, you’re back. What did you think? I listened to the sample of Agatha Christie’s Mysterious Affair at Styles. It was not as awful as I anticipated though some of the pauses seemed a little odd to me. I don’t know that I would have figured out that artificial intelligence was at work if I hadn’t known in advance. Still, I don’t think I want to listen to an entire book.

I’m not sure how I feel about this marriage of computer science and book content. A part of me is interested in how natural language processing has progressed since I was in college when it was fairly primitive. The part of me that consumes books is resistant to the idea.

On the plus side, for DeepZen anyway, they say that the voice over artists whose voices they use get paid for every audio book that is produced using their voice. Don’t know how much or how it relates to what the voice over artist would get if they read the audio book in the usual way. I suspect it’s less. On the other hand, they don’t have to sit down and record the audio book. Audiobooks can also be produced cheaper and faster using this method. But, from the little I’ve heard, the quality doesn’t seem as good as would result from a voice over artist reading the book. On the other hand, this may make some content available where it wouldn’t have been before.

There’s also the issue of rights. Do these need to change with the development of AI narration? I’m not a lawyer, I don’t know the answer to this, but I think the question needs to be asked.

While from a technology standpoint I find this fascinating, it still makes me uneasy. It seems like it’s taking away jobs from voice over artists. Plus I really prefer a good narrator for the audiobooks I listen to. But, then, I’m not dependent on audio for my book consumption. I can choose to read the book. Not everyone has this choice.

If you want to read more there are a lot of articles on the web on this topic. Here are a few of them: https://www.tckpublishing.com/ai-narration/

 https://towardsdatascience.com/how-ai-contributes-to-the-audiobook-industry-boom-2cc5406331eb

  https://www.thecreativepenn.com/2020/12/04/voice-technologies-streaming-and-subscription-audio-in-a-time-of-artificial-intelligence-ai/

So what do you think about the use of AI in audiobook narration?