• The Neuron
  • Posts
  • 🐈 One AI = unlimited voice actors?

🐈 One AI = unlimited voice actors?

PLUS: Another ChatGPT competitor, the AI talent immigration problem

Happy Friday morning! This is The Neuron, your daily AI multivitamin.

Welcome to the ~2,000 of you who joined this week! If you liked the newsletter, please do us a solid and share it on Twitter or forward this email to your buddy.

Today in AI:

  • This AI Makes Unlimited Voice Actors

  • Google DeepMind's ChatGPT Competitor Is Coming

  • Chart: US vs. China On AI Talent

  • Around the Horn

  • Leo Sends His Regards

This AI Makes Unlimited Voice Actors

Unlimited, high-quality voices.

Current text-to-speech tools all spit out either one human-like voice with tone/emotion or a bunch of robotic voices that clearly sound like computers.

Eleven Labs' new voice generator gets you both: unlimited human-like voices where you choose the gender, age, accent, pitch and speaking style.

Just press "Generate" (and "Generate" again if you don't like it) and have it read your text. Take a listen on their blog post.

Why it's important: Products like this and Play.HT make it cheaper to create voice tracks for audiobooks, news, games, ads, videos and more. And for anyone else who hates hearing their own voice on recording, it might be for you, too!

Stepping back: The full creative process is slowly starting to fill in with AI tools. For example, a 3D character can now speak using just AI tools to bring it to life.

Just type some text, choose a voice, then use NVIDIA's Audio2Face (which animates a 3D character's face based on the text you want it to say) to animate the face. Like this, but imagine a better AI voice:

Stacks of tools like these are a huge deal for industries like gaming, where it can take 4-6 years to launch a game.

Google DeepMind's ChatGPT Competitor Is Coming

ChatGPT, Claude, meet Sparrow...sometime in 2023.

TIME published its interview with DeepMind CEO Demis Hassabis today. It mostly talks about his story, artificial general intelligence (an AI that's as smart as a human) and how to do it ethically.

Buried in the interview was a hint about Sparrow, DeepMind's ChatGPT competitor: it's coming to private beta in 2023. It'll join Anthropic's Claude in the battle against ChatGPT for best chatbot.

From what I'm hearing, Sparrow has been ready for a while (it was first announced in Sept 2022) and DeepMind is just not very decisive about releasing it. Could be their concern around ethics.

Two other details from the interview:

  • Sparrow will be able to cite its sources, which neither ChatGPT nor Claude do.

  • DeepMind is wary of "freeloaders" and says "the AI industry’s culture of publishing its findings openly may soon need to end". That's like OpenAI, who has kept GPT-3, DALL-E and Whisper closed source.

Careful, now: DeepMind is a Google subsidiary, but you shouldn't treat this as Google's answer to ChatGPT. DeepMind is separate enough that there's a distinction.

Chart: US vs. The World On AI Talent

Don't drop the ball now, Uncle Sam.

This project tracks global AI researchers, and it made the chart above.

20% of top-tier AI researchers come from the US but 59% of them work in the US. Boom.

The employers? Those would be Google, Microsoft and Meta, along with every top research university you can think of.

This shouldn't be a surprise. Immigration is and will always be a strategic pillar for the US to win in a field like AI, just like it is across STEM. Let's just hope our lawmakers are paying attention.

Around the Horn

Leo Sends His Regards

That's all we have for today. See you cool cats on Twitter if you're there: @nonmayorpete

What'd you think of today's email?

Login or Subscribe to participate in polls.