Elevenlabs.io - AI Generate anyone's voice with a minute of audio saying how much they hate trannies

[Patrick Bait-man]

JC Denton reads part of Industrial Society and it's Future by Theodore John Kaczynski.

 

[MrJokerRager]

Video of Tucker doing a Star Wars style report on woke wookies.

 

[Burgerasssand]

This thing has been completely neutered into the ground. It was WAY more emotive in the early days but now it just sounds like a moderately improved text-to-speech.
Compare one of the first posts in this thread

This was posted on the sharty earlier today, and hoooooly shit this is bad.
View attachment 4368525

I would be surprised (but unimaginably thankful) if fake "discrete recordings" of public figures saying insanely unhinged shit didn't start making the rounds during this next election cycle. And keep in mind, this Eleven Labs thing is what's publicly out there - Lord only knows what's being (or already been) developed in private.

to

Video of Tucker doing a Star Wars style report on woke wookies.

 

[Xenomorphs Are Cute]

Microsoft made their own AI called VALL-E where only a 3 second clip is required. There's an unofficial version out.

 

[Rembrandt]

This thing has been completely neutered into the ground. It was WAY more emotive in the early days but now it just sounds like a moderately improved text-to-speech

After spending my entire quota during the first month of the craze, the 30 days have passed and yeah, it sounds worse. New cloned voices never sound remotely close unless they're female, the audio bitrate/compression varies substantially from attempt to attempt, it was an absolute drag to work with.
They successfully marketed it, then set up the paywall, reduced its quality so it's not as unrecognizable from real speech and are now making bank from normies who will just up and pay for it blindly after watching the memes. Fun has ended

 

[jspit2.0]

So the creator of Tortoise has dropped support for the Colab. His claim on the github is that it's google's fault, however this week he went on an autistic rant about how the tech was being used to scam people, and mocking the idea that anyone should be able to use it.

Looks like some other people are trying to maintain Tortoise without him, but we'll see how that goes.

Interesting to note is he seems to believe that Eleven Labs is using a further trained version of Tortoise to run their service. Not sure whether he's onto something or if he's just sneeding.

That's possible. Eleven Labs big thing is the way it conveys emotion I feel.

If this is true, that missing link between Tortoise and Eleven labs is the key.'

It'll be harder for this to make any more impact when there are text gens and audio gens you can easily run on consumer hardware like with stable diffusion. Text gen will take longer to get there but the push is there.

The issue is two fold when making AI material open source.

First, creating a friendlier interface with lower tech requirements

Second, dealing with changes to the law. It doesn't matter if open source happens if the law is altered to allow services and sources to begin wielding a version of DMCA. For example. If a voice actor can 'own' his voice, he could in theory pursue it's appropriation.

this will be realllllly interesting in the near future, to know that shit like anime dubs or hell even foreign shows being dubbed over with AI voices using the VAs' country of origin will be really fascinating to witness.

Now Cory in the House Anime will be within our grasp in no time.

I can't imagine how this will turn given how litigious the Japanese are. Just look at the Abridged phenomenon. Never a clearer instance of fair use and boom! DMCA'd to hell.

 

[Toolbox]

Have they made it difficult at all to generate content other ai services have deemed inappropriate? With chat you can at least jailbreak but this is just a service that takes text and turns it into audio. Is it worth it at this stage to go through account setup and purchase if they're just going to rug pull?

 

[Rembrandt]

Have they made it difficult at all to generate content other ai services have deemed inappropriate? With chat you can at least jailbreak but this is just a service that takes text and turns it into audio. Is it worth it at this stage to go through account setup and purchase if they're just going to rug pull?

Terms of service were updated a long time ago to say they have the right to notify the authorities if they find inappropriate stuff being generated. Since you must give away your credit card information to them and e-mail address etc. it wouldn't be hard to track you down. It didn't stop anyone from making offensive stuff and posting it on YT however, so not sure if it blocks prompts outright

 

[Toolbox]

Terms of service were updated a long time ago to say they have the right to notify the authorities if they find inappropriate stuff being generated. Since you must give away your credit card information to them and e-mail address etc. it wouldn't be hard to track you down. It didn't stop anyone from making offensive stuff and posting it on YT however, so not sure if it blocks prompts outright

I wonder what these niggers actually report. Is it in the UK context where they'll actually spill on anyone that generates audio with faggot in it? What would be considered illegally inappropriate in an audio generator? Making threats to politicians in the voice of Morgan Freeman?

 

[⠠⠠⠅⠑⠋⠋⠁⠇⠎ ⠠⠠⠊⠎ ⠠⠠⠁ ⠠⠠⠋⠁⠛]

Terms of service were updated a long time ago to say they have the right to notify the authorities if they find inappropriate stuff being generated. Since you must give away your credit card information to them and e-mail address etc. it wouldn't be hard to track you down. It didn't stop anyone from making offensive stuff and posting it on YT however, so not sure if it blocks prompts outright

Well, depends if they screen out prepaid debit cards.

After spending my entire quota during the first month of the craze, the 30 days have passed and yeah, it sounds worse. New cloned voices never sound remotely close unless they're female, the audio bitrate/compression varies substantially from attempt to attempt, it was an absolute drag to work with.
They successfully marketed it, then set up the paywall, reduced its quality so it's not as unrecognizable from real speech and are now making bank from normies who will just up and pay for it blindly after watching the memes. Fun has ended

This is capitalism in action. That's what you get from Ammurrica. You'd expect a paid account to be somewhat higher quality, though not the full quality- that'll be reserved for say, the National Endowment for Democracy or Mossad, and other criminal organizations that want to overthrow free governments.

I wouldn't be surprised if they're encoding a 25khz tone or something like that into regular free/paid account generated speech too, so they can sell 'ai voice detection' to retards.

 

[Sneeds]

Lil Kiwi

 

[Sneeds]

 

[102]

No idea how the creator of this one made the AI sing, first I've seen.

 

[Cowboy Kim]

Diff-SVC is used for singing (atleast in this uploaders case.)

 

[Patrick Bait-man]

 

[Maldavius Figtree]

Crazy how it took less than 24 hours for this to stop being funny at all.

I have to say with the presidents playing vidya clips it turned around and started being funny again

 

[Mysterious Autist XX]

I hope someone makes a video of several world leaders playing Payday 2

 

[Colon capital V]

I hope someone makes a video of several world leaders playing Payday 2

English Hitler arguing with English Stalin over why Gandhi should go nuclear in Civ would be pure kino.

 

[Mysterious Autist XX]

English Hitler arguing with English Stalin over why Gandhi should go nuclear in Civ would be pure kino.

Gandhi justifying to Hitler why he nuked Stalin would be funny now that you mention it.

 

[SuperShibuyaFighter]

This thing has been completely neutered into the ground. It was WAY more emotive in the early days but now it just sounds like a moderately improved text-to-speech.
Compare one of the first posts in this thread

to

After spending my entire quota during the first month of the craze, the 30 days have passed and yeah, it sounds worse. New cloned voices never sound remotely close unless they're female, the audio bitrate/compression varies substantially from attempt to attempt, it was an absolute drag to work with.
They successfully marketed it, then set up the paywall, reduced its quality so it's not as unrecognizable from real speech and are now making bank from normies who will just up and pay for it blindly after watching the memes. Fun has ended

I completely disagree. I have been testing it and have been getting great, realistic, and emotive sounding results.

Some voices it does better than others and sometimes it just depends on the samples, but listen to this test I did, there is a ton of emotion, especially near the end. My main issue overall is not being able to control the speed.