I really struggle with audio books that do that. It seems like it shifts some internal perception of mine from “book” to “audio drama” and ironically, it makes it much harder for me to get immersed in the book.
It’s weird. When there’s just one narrator and they do different voices for different characters I don’t think twice about it but I can still tell the characters apart by voice alone. But when there’s an entirely different person’s voice my attention gets mildly distracted by it. It’s similar to when sometimes books will have a random sentence read out by someone different than the narrator, probably a post recording correction.
That being said it’s usually pretty few and far between when I come across an audiobook book with multiple narrators, and usually it’s per section of the book (maybe from the perspective of different characters) rather than switching mid dialogue between characters
I tend to agree with you. The one exception I find is that when there is a single narrator, it seems bit off when they voice the opposite sex.
I’ve actually found myself surprised at how my brain doesn’t hear that was weird. I personally find it very natural. Maybe it’s because I was read a lot of stories verbally as a kid?
Same here. I still prefer single narrator. There are a few cases when there are just too many characters but it’s still much easier to listen to than multiple narrators.
I have also noticed sound effects in audiobooks. I like it at the end/start of a chapter, but it need to be subtle. I listened to Fractal Noise, which has audio effect for the thumbing sound, very quiet at the beginning but turned very loud at the end of the book. While it’s new and interesting at the beginning, I quickly grew tired of it. I’d rather only the narrator reading the book than hearing the sound effect.
I think it’s a matter of imagination. Reading/listening for me is not only about the story but also about my imagination. The sound effects removes this, sadly, despite the huge effort by the team.
Oh, what timing! Actually I’m working atm on recording an audio fiction for a big audiobook company. It has a fully voiced cast, and even sound effects.
I don’t know what details I can give because of NDA, but I assure you audio fictions are being made to this day.
Edit: and I guessed I missed the AI part, but I hope it doesn’t come soon cause I’ll be out of a job :'D
Super cool, how’d you get started? Always wanted to give it a try.
I was just lucky to get hired at a dubbing studio after finishing my studies 🙂
We mostly record dubs for movies and series, but we also record audiobooks or voices for videogames sometimes.
We’re doing this audio fiction thing for the first time. But maybe more will come.
deleted by creator
You could literally use AI to mark quotes for which character was supposed to be speaking it according to the context.
deleted by creator
Elevenlabs Projects seem to make them possible now.
As for AI voices, it doesn’t quite match real life voices yet. I think what’s more feasible is to get an above average actor with a decent budget, then mask the voice with programs like RVC.
Elevenlabs is incredible but yeah it’s not there 100% for audiobook purposes I think. I trained it on an audiobook from one of my favorite narrators and had it read from another book. It sounded just like the guy but the intonations and mannerisms just weren’t quite there to match the text.
I don’t even know of many non AI read audiobooks that use multiple speakers, or even a single speaker who gives different voices to different characters. I would absolutely love to see that more. I think the last audiobook I listened to where the reader did this was Neuromancer. And it annoyed me that the next two books had a different reader who didn’t give the voices I grew to like for the returning characters.
Frank Muller, who did some of the Dark Tower books, did a great job of differentiating character voices.
Also, the Dune books would occasionally use different voice actors for major characters, such as Baron Harkonnen. It wasn’t always - just when there was a scene that doing so really added to the atmosphere of the reading. Occasional sound effect too.
The books I’ve listened to from Tantor Audio and Random House Audio don’t always have multiple people reading, but the narrators are usually good and do distinct character voices.
I can second this. Perhaps not a distinct voice for every character, but any recurring character.
Speaking of single speaker audiobooks with multiple characters, I listened to an audiobook by Raza Jaffrey: The Stone Song. The characters were well acted, in my opinion. Some characters are kids, men, women, even an alien or two. But it’s easy to distinguish them. And the thing is, he didn’t even have to adjust his pitch. Goes to show how you need a specific skill if you want to do single speaker audiobooks with multiple characters right. Then again I am biased in a way that this is one of the first audiobooks I’ve ever listened to.
RC Bray does this very well in his narrations.
I’ve listened to the whole Witcher saga of audiobooks and the narrator is amazing. He does different voices for all characters, including women and children.
He’s not a voice prodigy like Mel Blanc or Robin Williams but he’s good enough that I had a difficult time accepting other audiobooks with less talented narrators after that.
The new Discworld audiobooks have a huge cast. Specific actors voice specific characters through the whole series.
Elsewhere, Travis Baldree always does awesome voices for characters in the audiobooks he reads.
Another question: Is there a word in english for the kinds of audiobooks that are a proper play? Acted out by different voice actors and with sound effects? Do you listen to those? In german you make a distinction between ‘Hörbuch’ and ‘Hörspiel’…
I guess we would call it a “radio play” because that sort of format has been around on radio for a long time. I’ve never come across one produced as an audio book, rather than for radio first, so I don’t know if anyone has tried to coin another name.
https://www.audible.com/pd/Lincoln-in-the-Bardo-Audiobook/B01MYV9WSI
Here you go. It’s amazing.
Full Cast Audiobook
Thanks. Are you from the UK? Because I’ve heard about BBC audio drama, Douglas Adams radio dramas etc…
Just out of curiosity: Do American people use that term, too? Do they listen to the radio and fiction/drama is part of radio in the US?
But thanks for all the terms. I think I can live with ‘radio play’, ‘drama’ and a ‘full-cast’ if I want to specify.
Yes, UK (sorry, should have said). And yes, the BBC commissions the vast majority of drama for radio. No idea if the US has any kind of equivalent.
The closest we have to radio plays and audio drama are podcasts. Except for some very niche stations, nobody is broadcasting fictional stories over American radio waves.
Talk radio now is more news and conversation. Not episodes of Dick Tracy or Little Orphan Annie like we had in the first half of the 1900s.
I only want this if it’s TikTok style AI voices of all the presidents teaming up to read the lorax
There is a company that does this with real voices now…
https://www.graphicaudio.net/our-productions/upcoming-releases.html
This will take quite some effort. I’m editing books, and sometimes I don’t know who is speaking a line (then I ask the author to clarify that point, but they don’t always do).
The real question is whether or not it is legal. Theoretically it is possible to do with current tech. If i was making such a tool, i would need access to the ebook then pass it through a llm model (possibly with a 7b open source one) to tag which characters are saying what. Once i have tagged dialogues then I could pass it through elevenlabs or other opensource tts and voila you have an audiobook with different voices.
The real problem is that opensource tts aren’t as good and i imagine if you use paid versions, you will encounter legal issues or it might be too expensive. And can you sell your audio book? Legal troubles again.
But if you just wanna do it while sailing the high seas, everything should be possible.
You can’t feed a book to an open-source LLM. I want to learn about something convenient that can take a book and generate an audiobook in just a few minutes, rather than a process where you have to divide the book into small parts to be fed into an open-source LLM and then produce audio from these parts and merge them together.
Not sure about AI, but there are a few audiobooks out there that feature an entire case, and sound effects.
