In partnership with

Good morning, human brains, and welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

• Text-to-Morgan Freeman in 60s 🗣️ ✍️

  ElevenLabs unveiled Voice Design.

• 2 AI hosts walk into a bar… 🤖 ❤️ 🤖

  Meta released NotebookLlama.

• Scrape any site without coding 📝 🤓

  A must-have Chrome extension.

New here? Subscribe! 😎

Peep today's ‘What Would You Do?’ at the bottom. 👇

MAIN COURSE

Your voice just got cloned 🗣️ ✍️

On Monday, ElevenLabs unveiled Voice Design. It’s a new tool that creates custom synthetic voices from text descriptions.

How does it work?

Write detailed voice descriptions like “old British male, raspy, deep voice, professional” and get a custom voice in seconds.

What can I use it for?

You can use it for indie game development, D&D character voices, student film projects, real-time voice generation for games, and more.

Can it replace voice actors?

Nah. While it’s useful for projects that can't afford voice talent, it can’t replace trained professionals… yet.

What makes a good prompt?

The more detailed the description, the better the result (though watch out for stereotypes).

What else has ElevenLabs done?

In June, we covered ElevenLabs’ Sound Effects. It allows you to create sound effects for your film, games, social media content, and more.

In July, we reported on ElevenLabs’ Iconic Voices feature. It enables you to have AI versions of Judy Garland, James Dean, and more read to you.

A couple of days later, we covered ElevenLabs’ Voice Isolator. It’s an app that eliminates background noise from your videos.

SPONSORED BY WRITER.

Writer RAG tool: build production-ready RAG apps in minutes

RAG in just a few lines of code? We’ve launched a predefined RAG tool on our developer platform, making it easy to bring your data into a Knowledge Graph and interact with it with AI. With a single API call, writer LLMs will intelligently call the RAG tool to chat with your data.

Integrated into Writer’s full-stack platform, it eliminates the need for complex vendor RAG setups, making it quick to build scalable, highly accurate AI workflows just by passing a graph ID of your data as a parameter to your RAG tool.

Learn more about our production ready RAG tooling here.

SIDE SALAD

Google's AI > Meta's knockoff 🤖 ❤️ 🤖

On Tuesday, Meta released NotebookLlama. It’s an open-source version of Google's NotebookLM podcast generator.

What does it do?

Turns text files like blogs or PDFs into podcast-style conversations using Llama models and text-to-speech.

How's the quality?

Pretty rough. The voices sound robotic and often talk over each other at weird times.

Can it be improved?

Better text-to-speech models could help, plus using two separate AI agents to debate topics.

How does it work?

It creates a transcript from uploaded files, adds dramatization and interruptions, and feeds to open text-to-speech models.

Any issues?

Like all AI, it still hallucinates (makes stuff up) in its content.

COOL TOOL 🛠️

FetchFox AI

FetchFox is a new way to do scraping. You can scrape any website with just plain English, no coding or complicated UI.

Tell FetchFox what you want to scrape: "Get the usernames of all the replies to this twitter post, and then go to their profiles and find their number of followers." With this command, FetchFox will scrape it for you.

Anyone can use the Chrome extension, and if you're a developer, there's also a free open source library that you can use in your own apps.

Get the extension.

YOUR DAILY MUNCH

Think Piece 🧠

AI Agent case study. Claude demonstrates impressive autonomy and strategic planning, but highlights AI’s limitations of reliability and depth in agent-based systems.

Can Hong Kong balance AI innovation and regulation in finance? It introduced a dual-track policy to promote AI adoption in the financial sector while addressing related risks.

Startup News 💰

Meta and Reuters announced a partnership. Meta will integrate Reuters' news content into its AI chatbot, enabling it to provide news-related answers with citations.

Research 👨‍🔬

ROCKET-1 — a policy model that utilizes visual-temporal context prompting to enhance vision-language models for open-world interactions.

Infinity-MM — a 40 million-sample multimodal instruction dataset enhanced with quality filtering and deduplication.

FURRY FRIENDS

Say “hi” to Conan.

Respond to this email with your pet’s name and pic for a chance to be featured here tomorrow!

AI ART SHOW

WHAT WOULD YOU DO?

Horses are overrated. 🐴

What giant animal are you riding? 👇

Ideas? Comments? Complaints?

Respond to this email or hit me up on 𝕏.

Until next time 🤖😋🧠