dont even have enough professional voice actors to cover the bandwidth
I’m pretty sure they’d be a lot more people ready to do that job if there was a good remuneration. Heck that sounds a lot more fun that a LOT of jobs out there!
Comment on Audible unveils plans to use AI voices to narrate audiobooks
drmoose@lemmy.world 3 days ago
This is clearly the future despite the outrage here.
There are at least 389 living languages with over 1M speakers. That alone means it’s impossible to reach some people and they get left out. Most of these languages dont even have enough professional voice actors to cover the bandwidth.
There are thousands of books released every year. That’s impossible to cover even in English alone.
Its an objective net good to have more accessible audio books and the privileged people who do care about this stuff can very much afford to vote with their wallets for non-ai voices.
utopiah@lemmy.world 3 days ago
drmoose@lemmy.world 3 days ago
Sure but that’s not how free markets work. If there’s only 3 million consumers you can’t afford 3 million voice actors but you can afford 3 million AI renders.
utopiah@lemmy.world 2 days ago
I’m not an economist but… 1 voice actor can serve 3 million consumers if they listen to the same content.
Anyway that’s not even my point, my point is that it is possible to cover, we as a society, driven both by VC with strategies of capturing markets (so precisely going against “free” market as an ideal) and consumers are making choices (like when one buys from the local farmer market vs Amazon deliveries). If though we, while fully understanding the consequence of such choice (namely how the sausage is made, here how AI models are trained and then run), believe it’s not valuable then sure, we can make that choice.
I’m just warning consumers then that if they don’t pay for quality content made a certain way, they can’t complain that they in turn don’t get the job they wanted because nobody out there is ready to pay for it.
2 sides of the same coin.
DNU@lemmy.world 3 days ago
but for a service like audible.
taladar@sh.itjust.works 3 days ago
And you think anyone is training AI voice models for those languages? Have you even seen how long it takes even large companies like Google to support the languages with hundreds of millions of speakers?
drmoose@lemmy.world 3 days ago
It becomes easier and cheaper every day. Today’s open source LLMs are better than last year’s best model.
ExperiencedWinter@lemmy.world 3 days ago
You’re fundamentally misunderstanding the comment you replied to, they are not saying that voice AI are bad, they are saying there is not enough training data to improve the AI for these languages. How will it improve without good training data?
drmoose@lemmy.world 3 days ago
Thats not how AI training works and even then there’s absolutely enough data. Also training data can be created and even synthesized. There are many techniques to extract make training value from datasets that we discover every year - It’s really not a problem you think it is.
I’m genuinely confused how AI illiterate users here are. It’s just blind leading the blind.
Jhex@lemmy.world 3 days ago
Is it? I just tried again yesterday for a simple script since coding is the one thing apparently AI will replace people like me and it could not put together a working JavaScript script.
I have yet to see tangible results not announced by the people with sunken cost exploding their balls.
drmoose@lemmy.world 3 days ago
Sounds like a skill issue my dude. Whole you struggle to get a js script people are putting out entire programs with AI assistants so sure - you’re right and they’re wrong
JohnEdwa@sopuli.xyz 3 days ago
That’s the benefit of using AI and machine learning - once you have enough source material, you can throw it all in and it’ll eventually spit out a model.
Which is exactly what Meta did with their Massively Multilingual Speech project which supports text -to-speech and speech-to-text for 1107 different languages.
Is it actually any good in 99% of them, I don’t have a clue, but it exists.
taladar@sh.itjust.works 3 days ago
Seems more like a proof of concept project for that paper than something they are pursuing seriously judging by the GitHub location in some example folder that hasn’t seen any significant updates in over a year. If it is so great I would assume they would pursue it more actively and replace existing models with it two years later.