I just listened to this AI generated audiobook and if it didn’t say it was AI, I’d have thought it was human-made. It has different voices, dramatization, sound effects… The last I’d heard about this tech was a post saying Stephen Fry’s voice was stolen and replicated by AI. But since then, nothing, even though it’s clearly advanced incredibly fast. You’d expect more buzz for something that went from detectable as AI to indistinguishable from humans so quickly. How is it that no one is talking about AI generated audiobooks and their rapid improvement? This seems like a huge deal to me.
A lot of people just aren’t aware of how fast AI is moving. AI voices were pretty meh earlier this year. A lot of people working on the audiobook/voice acting scene have been talking about this though.
I recommend everyone to check the YouTube channel “two minute papers” who have being doing videos about papers on AI for the last 10 years on so to see the accelerated progress AI have. Like 5 years ago those images generating AI looked like LSD infused dreams and now they look almost perfect.
I wish I could watch his videos but the way he talks is awful. It’s like some exaggerated evolution of YouTube talk.
I’m only shocked that video isn’t better. Diffusion models work like denoising - so you’d figure all the wiggly nonsense between frames would be the first thing to filter out.
I expect the data size to be a problem. Stable diffusion defaults to 512x512px, because it simply requires a lot of resources to generate an image. Even more so to train one. Now do that times 30 to generate even one second of video. I think we need something that scales better.
I fully expect this to work decently in a few years though, no matter how hard the challenge is, ai is moving really fast.
I give it a year, maybe two, for a fully synthetic video that couldn’t not be easily distinguish from reality. There’s already some very good AI that complete or replace backgrounds on videos that work really good, and completely synthetic videos that looks like nightmares for now.
Ah yes, Audio AI. I can’t wait for this rapidly-approaching future where you literally won’t be able to trust the validity of anything your senses tell you anymore
But up until this point, you see, there has always been one medium that is difficult/expensive enough to convincingly fake that it can reasonably be used as proof that something actually happened. If technology advances to the point where a video of something happening is no more convincing than a text description that it happened, and no other more sophisticated, harder-to-fake medium steps in to replace it…
I don’t want to live in a world where the truth is anything you can convince your friends of, you feel me?
“Up until this point” meaning maybe eighty years where unexpected events had any chance of being on film or televised, and several decades where amateur video was even theoretically possible.
And solid corroborating evidence still barely moved the needle whenever it was footage of cops trying to kill someone.
And what’s going to make bodycams necessary regardless is chain-of-custody demonstrating (a) the footage matching what the victims said absolutely came from the camera strapped to the chest of the accused, or (b) some motherfucker orchestrated a cover-up that demonstrates consciousness of guilt.
Imagine the day when people post videos of the president saying literally anything with pitch perfect audio voice synth
Imagine going to prison for a generated clip of you confessing to a crime.
Once the tech is that good, a recording of your confession will be useless as evidence in court.
…but it is already that good? The fact that celebrities are having to come out and say it wasn’t them in an ad is proof enough that it can fool people
You only need to fool a jury
That got me thinking about when we’ll hear the first case of AI generated security camera footage used to frame someone. Which leads me to wonder when it will be standard procedure for cameras to digitally sign their footage.
Everything will be useless in court. Audio evidence? Worthless. Video evidence? Worthless. Physical evidence? Prove that it wasnt planted. That kind of AI is a fucking nightmare and no one really understands the danger that kind of AI poses.
You know some people are just gonna generate that fucking locker room smell, the reek of hormones and axe body spray, to terrorize people
Tech like this has been available for a number of years, and has most likely already been used against you. It’s now getting available for the broader masses, but that might just be a blessing in disguise, since increased awareness will hopefully also make you suspicious of those cases that are already happening.
Have you watched a movie, ever? There have always been special effects trickery.
Yes, but you could tell they weren’t real. They still needed real voice actors, real sound design, studios and stages and resources. Anyone with a halfway decent rig can fake shit to a very believable degree. Even with CGI you swear is fantastic, you see its fakeness once the novelty wears off
I guess that AI generated stuff will also have some telltale signs of being fake for quite some time if you actually look for it.
If your phone is rendering TTS on the fly that’s probably going to be a drain on battery.
As someone who only consumes books in audiobook form this is great news for me, I tried to listen to some automatically generated audio books around 2 years ago and I found them horrible to listen to just because they sounded so off.
I’d love to be able to copy in the text of a book and get actually listenable (is that a proper word?) audiobook out of the other side for some books that will just simply never be recorded by actual people due to being too old / obscure.
I’ve been wanting to be able to listen to the Pelucidar books for years but they just don’t exist in audio format, is there somewhere publically available that I can do this?
I would guess there is a LOT of work going into each voice. Playing with different parameters and prompts. I don’t think it’s as simple as just copying the text into a box. Not yet at least :)
That’s a good thought there though. Audiobooks could have each character voiced uniquely.
This is literally the only upside I see from this.
One of the Dune audio books started off as multiple voices and then part way through it was finished by just one guy. Really impressed with it at first, and then really kind of debuffed by it. I had already read the book years before so it wasn’t a big deal, but like wtf?
I can hope! With the speed things are developing it may not be too long.
I haven’t played around with or looked into much to do with AI at all but would be willing to put in some time into playing with prompts / parameters if it meant I could eventually create a reliable work flow to create things such as what I mentioned.
I think I’ll have to do some research, I need some more old school hollow earth stories in my life xD
Just curious, but how come you only consume books in audio format? (Please forgive me if this was rude to ask.)
I can’t speak for OP but I do this as well. For me it’s because I listen to them on the drive to/from dropping my kids off at school and I’ll have it playing while I’m working or playing a game.
As someone who would like to do this, how well do you actually pay attention to what is going on? I’d do so much more reading if I didn’t have to go back and reread paragraphs several times over because I simply can’t pay attention, let alone if I’m doing something else entirely
Not OP, but I almost exclusively read novels and non fiction via audiobooks. For context, I’m on pace for 70 books this year.
My main reason for audiobooks is I having a driving commute. Two hours a day round trip. Audiobooks keep me sane in a way that podcasts or music do not. I also do audiobooks when doing chores around the house.
Second, I struggle to focus on reading a book on my phone. Too many distractions and I think the reading experience is subpar. I do have an eInk reader, but I haven’t charged it in years because it’s easier to do audiobooks.
Physical books are rare in my home, but that’s a self-reinforcing cycle since I enjoy audiobooks so much.
I like to read books before bed, but need darkness for a while before I have any chance of going to sleep, so me and my wife listen to 45min of audio book a night before going to sleep. Plus when we listen together there is no need to worry about getting ahead of each other and spoiling stuff.
I read books in other scenarios but that ritual is by the most time I have for reading and the most consistent as well.
Not rude at all, similar to the other responses people have given but it oa two fold really. Firstly I just don’t do well with sitting and reading a book, I get bored very quickly, can’t concentrate on what is happening and start re-reading sentences or pages over and over where I am not paying attention properly. Additionally after only a couple of pages it will start putting me to sleep, I guess my attention span is just not sufficient for this form of media.
As a result I never read any books until I discovered audiobooks and my love for them, I honestly just disregarded books as a form of entertainment and thought they were a waste of time until discovering this way to consumer them which wasn’t until I was in my early 30s.
On top of that I now listen to them mostly at work, I work with industrial machines and the work is repetitive as fuck and having a book to listen to makes the time go a lot faster and in a lot more interesting manner. Consequently I now love books and will listen to between 6 and 10 hours a day and now listen to them when I’m doing things like cooking, cleaning or running when I am not at work.
Back in the 19th century when unions were powerful and innovative, a lot of people had jobs where they had to sit and do repetitive tasks in a room all day. A lot of it was handwork that didn’t have big loud machines.
So one of the demands made by workers in such situations was that the employer would pay someone to come in and provide entertainment such as reading a book or giving talks on subjects of interest. The book or lecturer of course being selected by the workers via the democratic process of the union. And then of course the workers became way more educated because they suddenly had 8-12 hours daily to read books together. Since knowledge is power, the workers became stronger and more decisive in their collective actions.
When you are listening to audiobook at work you can know you are in a long tradition of workers exercising power over their job conditions. Although now it is individualized in the implementation. The desire to have your mind even though the job has your body and some concentration is universal.
Personally I mostly use audio books instead of reading because I get eye strain a lot easier than I used to. I go to an eye specialist for unrelated issues yearly, so it’s not an issue with a wrong lens prescription. It’s not a problem when I’m doing a low attention task where I can look away frequently, but for reading it sucks.
Well you can always pay someone to read it for you. Blind people do that.
Are any of these books public domain? If so the print version could be eligible for inclusion at Project Guttenberg. PG has very specific docs about eligibility for this. You could probably get a scan from archive.org if you don’t have one. You would have to clean up the OCR by hand.
Then it would eligible to be requested from the volunteer (human) readers who have been pumping out Libra audio books for years at LibriVox.
Recently I saw Gutenberg has a collab. They are producing and distributing Libre guidebooks generated by AI. I believe I read on one of the pages they have 4000 done. I haven’t tried it out but I guess I should.
Project Gutenberg, Microsoft, and MIT have worked together to create thousands of free and open audiobooks using new neural text-to-speech technology and Project Gutenberg’s large open-access collection of e-books. This project aims to make literature more accessible to (audio)book-lovers everywhere and democratize access to high quality audiobooks. Whether you are learning to read, looking for inclusive reading technology, or about to head out on a long drive, we hope you enjoy this audiobook collection.
I assume this is also a great benefit as fertilizer down at the old AI content farm which is otherwise totally run over with reddit shitposts.
If anyone tries it let me know how it goes.
The books I specifically mentioned are now public domain as they are old enough and librevox is where I actually started my audiobook (and books in general) journey. One of them is on there but it is only the second book of what is a 5 or more book series which is kinda frustrating.
The volunteer readers are very hit and miss however and I find that more than half are just not listenable for me due to different reasons from poor actual recordings, poor reading ability by the reader with excessive pauses added “errs and ummms” to mispronunciation of words constantly. These are pedantic reasons maybe and I throw no shade over it to the people that have volunteered their time to read these books but I just can’t listen to them personally for the same reason I could never get through any amount of time with a robotic text to speech program of the past.
I’ll look into the project Gutenberg thing however, thanks for making me aware of it and see what is up with that :)
Totally true about the librivox readers. They are doing their best. :) There are some total gems in there. But I have definitely given up on a few of them. OTOH I have given up on professionally read audiobooks too for all sorts of reasons.
I’m pretty sure that Amazon tried to do this with Kindle a few years ago and got sued by book publishers.
Ahh, it was audible.
It’s only a matter of time though before this sort of thing is ruled on and deals are inked. Open source is already getting pretty far too.
I listen to a lot of audiobooks too, but I wouldn’t listen to something like this.
Have you listened to the one OP posted? After a minute I’m sleeping. There’s no emotion, no tension, nothing.
I can’t stress enough how bad OP’s sounds. Sure, it sounds natural when compared to what technology was capable of some time ago, but it’s dead inside.
Good voice actors bring a book alive. This doesn’t.
I hadn’t actually when I originally wrote this comment as I wasn’t somewhere that I could listen to audio (without being an obnoxious cunt and I won’t be one of those people even if the majority of people seem to be OK with it xD)
However I would agree, I couldn’t listen to a whole book with this monotonous drone of a voice, however like you said compared to what was produced by voice to text type systems of the past this is miles ahead and I certainly look forward to the technology progressing to a point where it is listenable for me.
I agree with good voice actors being essential for enjoyment, I think that is one of the reasons I fell in love with graphic audio productions over the last couple of years.
Look at the description of the video. It’s not automatically generated. He made several voices and narrator and applied it to each character.
While insanely cool, it’s not “put in book here, get audio book there”
Yes I realise that and was over simplifying in this response but as I stated in another comment I would be more than happy to work on prompts for myself if it could generate something satisfactory to listen to.
The video posted by OP still sounds a bit “dead” so I don’t think the tech is quite there yet but it is promising for the future the way it is headed.
That sounds pretty cool, though I’d be concerned it will suffer from the classic problem of current AI (…and humans, but that’s by the by) of confident incorrectness. Like an automatic transmission can miss meanings and types of context that a human will spot, programmatically generating speech can probably mess up punctuation and flow - even the way a human reader sometimes will get part way through a sentence and realise they need to start again for it to come out right.
That said, I can’t see it being a big problem for most works, just unfortunate here and there. For once it seems an AI application short on downsides! (Except for the usual economic ones for many people previously trained in the field.)
