X

The voice-generating tools from ElevenLabs are now available

With support for more than 30 languages, ElevenLabs, the popular AI-powered platform for producing synthetic voices, today debuted its platform.

ElevenLabs says that its technologies are now capable of automatically recognising languages, such as Korean, Dutch, and Vietnamese, and producing “emotionally rich” speech in those languages using a new AI model created in-house.

Customers of ElevenLabs can use the voice-cloning feature of the platform together with the new model to talk in approximately 30 different languages without having to first input text.

ElevenLabs CEO and co-founder Mati Staniszewski stated in a statement: “ElevenLabs was started with the dream of making all content universally accessible in any language and in any voice.” “With this release, we’re one step closer to making this dream a reality and making human-quality AI voices available in every dialect. Our text-to-speech generation tools help level the playing field and bring top quality spoken audio capabilities to all the creators out there.”

ElevenLabs, which was founded by former Palantir employee Staniszewski and his childhood friend Piotr Dabkowski, a former Google employee, has recently made news for both admirable and repugnant reasons. As children in Poland, Staniszewski and Dabkowski watched American films with subpar dubbing, so they set out to create a platform that could do better—of course using AI.

Due to the high standard of its created voices and generous free tier, ElevenLabs gained traction shortly after its beta launch in late January. But as was already mentioned, the PR hasn’t always been good, especially once dishonest people started abusing the site.

The infamous message board 4chan, known for its conspiratorial material, shared abusive remarks using ElevenLabs’ tools while imitating celebrities like the actor Emma Watson. James Vincent of The Verge was also able to use ElevenLabs to quickly clone targets’ voices, creating audio samples that ranged from violent threats to statements of racism and transphobia.

ElevenLabs responded by announcing that it would implement a number of new security measures, such as limiting voice cloning to paying accounts and offering a new AI detection tool.

The threat that ElevenLabs’ platform and others like it pose to the voice acting industry is another debate that is now roiling around them.

According to a report by Motherboard, customers are increasingly asking voice performers to sign away the rights to their voices so that AI can create synthetic versions that may one day take their place. Meanwhile, Activision Blizzard, one of the biggest game publishers in the world, is reportedly developing tools for AI-assisted “voice cloning,” according to internal emails obtained by The New York Times.

With praise for its partnerships with publishers like Storytel, media companies like TheSoul Publishing and MNTN for audiobooks and radio content, and publishers like Embark Studios and Paradox Interactive for video games (Storytel and TheSoul Publishing are strategic investors), it would seem that ElevenLabs sees this as the natural course of events. According to the company, it has more than a million registered users from the publishing, entertainment, and creative industries who have produced audio material over a ten-year period.

ElevenLabs, which recently raised $19 million at a valuation of $99 from investors including Andreessen Horowitz and DeepMind co-founder Mustafa Suleyman, intends to eventually apply its AI models to voice dubbing, following in the footsteps of startups like Papercup and Deepdub and creating what it refers to as “a foundation to be able to transfer emotions and intonation from one language to another.”

In addition, ElevenLabs says it will eventually add a feature that will let users share their voices on the site, but specifics are yet hazy.

 

Categories: Technology
Priyanka Patil:

This website uses cookies.