Good morning. It’s Wednesday, June 19th.

Did you know: On this day in 1963, Soviet cosmonaut Valentina Tereshkova, the first woman to travel in space, returned to Earth in the spacecraft Vostok 6?

In today’s email:

• AI For Soundtracks, Dialogue

• China’s AI Sexbots

• Meta’s New Suite of Models

• 10 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

Get our book: Decoding AI |  Upgrade to ad-free

^{In partnership with WEGIC}

Introducing Wegic: Your AI Web Designer & Developer

Create stunning websites effortlessly with Wegic, the first AI-powered web designer and developer by your side. With Wegic, you can bring your dream website to life in just 90 seconds through simple chats.

🏆 Proudly named Product of the Day, Week, and Month on Product Hunt!

Why Wegic?

• Fast and Easy: Build your website in 90 seconds

• User-Friendly: No coding skills required

• Time-Saving: Ideal for indie developers, small business owners, and startup CEOs

DeepMind's New AI Generates Soundtracks, Dialogue For Videos

DeepMind introduces V2A, an AI sound director for silent videos. V2A, short for "video-to-audio," tackles the limitation of muted outputs in current video generation models. . This technology can create music, sound effects, and dialogue synchronized with video content, enhancing the realism of AI-generated media. Trained on a meticulously curated dataset combining audio, dialogue scripts, and video clips, V2A learns to sonically mirror on-screen action.

While V2A shows promise, it is not yet perfect; its audio generation quality can suffer in videos with artifacts or distortions. Due to potential misuse, DeepMind has decided against immediate public release, opting for further safety assessments and gathering feedback from creators and filmmakers. V2A aims to be a valuable tool for archivists and those working with historical footage, while raising important considerations about the impact of generative AI on the film and TV industry. Read more.

Related story: Runway Gen-3 Alpha: New video model closes gap with OpenAI's Sora

China’s Next-Gen Sexbots Powered by AI are About To Hit Shelves

Chinese manufacturer Starpery Technology are injecting AI into the "oldest profession," aiming for interactive dolls that converse and, well, connect. These next-generation sexbots, set to debut soon, promise enhanced user experiences by focusing on emotional connections through sophisticated AI models. Starpery's approach involves training LLMs similar to ChatGPT, enabling these dolls to respond intelligently and perform tasks beyond basic dialogue.

Despite technical challenges in achieving lifelike human interaction, such as realistic movement and speech, Starpery remains committed to innovation. Moreover, they plan to expand into robotics for household chores and care by 2025, leveraging AI advancements to address societal needs. However, the industry faces ethical concerns regarding consent, gender stereotypes, and privacy issues due to the sensitive nature of AI-driven intimate companions. Read more.

Meta Releases Flurry of New AI Models for Audio, Text, Watermarking

Meta's FAIR team has introduced a series of new AI models and tools focused on audio synthesis, text-image interactions, and watermarking. Leading the lineup is JASCO, an innovative AI that refines musical elements like chords and beats using textual inputs, allowing users to adjust generated sounds through simple text commands. FAIR plans to release JASCO's inference code and pre-trained models under open-source licenses to encourage collaboration in auditory AI development.

Additionally, Meta has launched AudioSeal, a tool for applying digital watermarks to AI-generated speech, enhancing the ability to detect synthetic content within real-world audio recordings. This advancement is aimed at improving detection speeds for AI authenticity verification.

Meta's Chameleon model, designed for tasks like image captioning using text, is currently accessible only for text-related functionalities. These releases underscore Meta's commitment to advancing AI capabilities while promoting transparency and innovation in the field. Read more.

Related story: Meta forms new Wearables group and lays off some employees

Etcetera: Stories you may have missed

AI shows how field crops develop: Software can simulate future growth based on a single initial image

Camb AI aims to challenge ElevenLabs - and releases its voice cloning model for free

TikTok ads will now include AI avatars of creators and stock actors

Google DeepMind Shifts From Research Lab to AI Product Factory

Kai-Fu Lee says the cost of AI inference compute will reduce by 100x in the next 2 years

OpenVLA is an open-source generalist robotics model

Nvidia eclipses Microsoft as world's most valuable company

OpenAI Expands Healthcare Push With Color Health’s Cancer Copilot

DeepSeek-Coder-V2: Open-source model beats GPT-4 and Claude Opus

Apple suspends work on next high-end headset, to focus on cheaper model 

Apple Developer Academy adds AI training for students and alumni

Notion launched 'AI connectors', a new feature that connects Slack and Google Drive to Notion AI

Epipheo AI is a free, generative AI tool designed to create professional marketing videos quickly, featuring dynamic visuals, compelling scripts, and voiceovers.

Vizly is an AI-powered data scientist that lets you chat with your data, visualize insights, and perform analysis.

Flyhomes AI is the first AI-powered home search portal, offering comprehensive, conversational real estate search and research without sales pressure.

ServiceSim is an AI-driven training simulator that replicates customer interactions, enabling agents to improve their skills beyond classroom training.

Agent Mode in Warp AI enables developers to perform tasks using natural language in their terminal, executing commands with permission and self-correcting capabilities.

Hamming Prompt Optimizer enhances task-related prompts using Large Language Models, generating optimized, structured prompts for fields like data analysis, healthcare, and finance.

Callin.io is the first AI phone assistant for small businesses, enabling rapid setup and seamless integration to enhance customer service efficiency.

GenType is an AI tool using Google’s Imagen 2 to create custom alphabets from any concept by generating 26 styled letters per user prompt.

Jamahook Sound Assistant is an AI-powered plugin for DAWs, helping music producers find harmonically and rhythmically compatible sounds from local and cloud libraries.

LinkRobot is an AI-powered tool that automates internal linking on websites, enhancing SEO and site navigation by suggesting and inserting relevant links.

📄 Make It Count: Text-to-Image Generation with an Accurate Number of Objects 

📄 DataComp-LM: In search of the next generation of training sets for language models ​​

📄 L4GM: Large 4D Gaussian Reconstruction Model

📄 Vid3D: Synthesis of Dynamic 3D Scenes using 2D Video Diffusion  

📄 mDPO: Conditional Preference Optimization for Multimodal Large Language Models 

📄 Transcendence: Generative Models Can Outperform The Experts That Train Them

📄 JEN-1 DreamStyler: Customized Musical Concept Learning via Pivotal Parameters Tuning

📄 ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools  

📄 VoCo-LLaMA: Towards Vision Compression with Large Language Models

📄OlympicArena: Benchmarking Multi-discipline Cognitive Reasoning for Superintelligent AI 

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, reply to this email with your inquiry for availability and pricing.