Read Time: 6 Minutes
Earlier this week I received and email from Kindle Direct Publishing (an Amazon service that you can self-publish through) which I have used in some dim and distant past to publish books and short stories I wrote under my pen name of Morgan Hobbes.
The email was about a new service they had to add “Virtual Voice” narration to my books. Curious if this was any good, I’d thought I’d give it a go and let you know the results. I thought I’d try it on “The Redaction of Vernon Reid” – a humorous short story I wrote and published back in 2018.
A Little Background
I haven’t listened to many books using Virtual Voice, but those I had I wasn’t too keen on. Obviously, I listen to a lot of audiobooks as demonstrated by this website where I review them, plus I have few audiobooks of my own done by actual narrators, so for me narration quality is quite a big deal. If it doesn’t sound right, then that is going to impact my enjoyment of the story.
The only book I have ever listen to using Virtual Voice was a collection of short stories in the Hard Luck Hank universe, which I wasn’t impressed with. These were published about a year ago (at time of writing), so I wasn’t holding my breath over much in the way of improvements. In my review, these were my comments about the audio quality:
These were not well done. First off, the quality was surprisingly bad. I had expected a clean, crisp sound but it sounded low quality. Most of all, I couldn’t stand the lack of any emotion behind the read. These are humorous books and need the comic timing that an AI doesn’t have. Further, there was no voice distinction between any of the characters and I found myself trying to keep up half the time with who was who.
One thing that’s often immediately noticeable in audiobooks by actual narrators is when they’ve spliced in retakes where their setup has obviously changed. You’d think that a computer reading a book would produce an identical output all the way through. Not so. For whatever reason, when do-overs (presumably for mispronunciations by the AI) were added in, the change in audio quality was very noticeable.
Would a year and change make a difference in quality? Read on to find out!
Production Effort
Actually producing the audiobook is a piece of cake. After logging into Kindle Direct Publishing (KDP) there is a link to add an audiobook with virtual voice. Clicking this link give you a few options starting with some pre-selected voices of American and British English in either masculine or feminine styles. There are others to choose from, but I found the American voice pleasant, so I went with that.
The actual production is done in the Virtual Voice Studio where you can listen to the voice reading the book, and make adjustments as necessary.
What I Liked
Each chapter can have its own voice, so if you’ve got multiple viewpoints or any other need to change voices, you can select a different virtual voice as needed. I didn’t use this function but it’s handy to have.
You can also add pauses, change pronunciation of words, and change the speed of the voice. An interesting “hack” I found was when, after listening to its say the chapter title, that I wished it actually said “Chapter xx” before saying the chapter name.
That’s when it hit me.
Using the pronunciation option, I could highlight the title (in the example “A Bomb Named Keith”) and tell it to pronounce it like “Chapter 1: A Bomb Named Keith” – essentially adding it a couple of extra words.
The pronunciation option I found particularly useful as I have a few weird alien names, so I could make it pronounce them phonetically. It would also identify all occurrences of the word throughout the story, so I only had to tell it once.
Audio Quality Sample
The quality of the actual voice is rather good. There aren’t any weird cuts or changes in volume or tone along the way. I’ve included a sample where the pronunciation change was used, and you can see it reads it without any apparent change in quality. Obviously, it’s clear of any background / studio noises which occasionally slip in with human narrators.
What It’s Missing
The main thing for me is how flat the reading is. Everything is read in a tone with no emotion or emphasis on any of the words. It does do an okay job at pausing where it needs to, but that’s about it.
If I had one upgrade suggestion it would be add an option where you could change the emphasis or inflection on a word. For example, I have a sentence where the bomb is talking and says:
“Not ‘the bomb’, ‘a bomb’. A Timeline Redaction Bomb to be precise.”
When read by the virtual voice, it reads as a single run on sentence like “not-the-bomb-a-bomb” … whereas with the quotes around ‘the bomb’ and ‘a bomb’ there should be emphasis on these, so it would read as “Not the bomb, a bomb.”
Overall Thoughts
It’s definitely an improvement over the other example of Virtual Voice I have listened to. Is it perfect? No, mainly because it’s missing the ability to put emotion and emphasis into the text it is reading. However, as far as quality goes, it has definitely improved. Changing the pronunciation of words doesn’t sound spliced in so the production quality is good.
As a cheap way to produce audiobooks I guess it works, but until it can replicate an actual human narrator being able to put soul behind the words, it’ll still be second place. Given the advancements in AI and technology in general, realistically that may not be too far away.
Plus, I’m not a fan of taking work away from humans. There are hundreds of great narrators out there and it would be a shame to see them replaced by automation.
I could see this being particularly useful for books where no narration exists and people need to add it for accessibility. Being able to add it to textbooks or instruction manuals, for example, would be useful. Even out-of-print or older books where the copyright owner or publisher can’t see a return on the investment of spending money to get an audiobook produced. I know there are a few books I’d love to have as audiobooks but they’re not available, presumably because of this.
If I had the option of making a personal “virtual voice” audiobook copy of a book I owned, I’d certainly take it over not having one at all.
From an availability point of view, the audiobook is automatically part of the Audible Plus range so is available for free as part of your subscription. If you’d like to see what Virtual Voice is like, why not check out The Redaction of Vernon Reid and see what you think! I’d love to hear your thoughts, so encourage you to leave your comments below.