Welcome back!

Several new audio and video tools are emerging. ElevenLabs launched Sound Effects AI, which transforms text prompts and videos into custom sound effects, offering new possibilities to creators. Let's unpack this...

Today’s Summary:

• ElevenLabs Sound Effects AI

• Humanoid robots with expressive faces

• Nvidia Hydra-MDP wins self-driving challenge

• Gemini new feature reduces API costs for developers

• Execs incorporate AI in business operations

• New releases from Meta FAIR

• Softbank AI to cancel angry customer voices

• 3 new tools

TOP STORY

ElevenLabs Launches Sound Effects AI

The Summary: ElevenLabs, known for AI voice generation, has launched its Sound Effects AI. This tool uses text prompts to generate up to 22 seconds of custom sound effects. 

Trained on sound effects from the Shutterstock audio library, it also features a new Video to Sound Effects tool that can import video clips and generate multiple audio options matching the visuals.

Key details:

• Free tier available with attribution

• Paid tiers for commercial use

• Prompt character limits for free and paid plans

Why it matters: Creators often struggle to find the perfect sound effect. Sound Effects AI offers a simple solution to generate custom audio from text prompts, making it an effective tool for creators.

ROBOTICS

Ex-Robots Humanoids with Enhanced Facial Movement

The Summary: A Chinese startup, Ex-Robots, is developing humanoid robots capable of imitating facial expressions. Using AI, they learn and replicate expressions through tiny motors. The company aims to integrate these robots in healthcare and education.

Key details:

• Takes 2 weeks to 1 month to produce one robot

• Cost between $207,000 - $280,000 each

• Aims for future uses in healthcare, education, services

Why it matters: Humanoid robots that mimic facial emotions can allow more natural communication. While current versions might seem uncanny, the technology is rapidly evolving. Future personal robots could either imitate humans or adopt more abstract functional designs.

SELF-DRIVING

Nvidia Wins Self-Driving Challenge with Hydra MDP

The Summary: Nvidia Research topped the CVPR 2024 Grand Challenge for End-to-End Autonomous Driving with Hydra-MDP. This model integrates human and rule-based knowledge, enabling robust performance in complex scenarios.

Key details:

• Hydra-MDP won CVPR End-to-End Driving at Scale challenge, outperforming 400+ entries

• Uses novel multi-teacher knowledge distillation combining human and rule-based data

• Multimodal inputs (camera, lidar) and multi-target planning for safety, comfort, efficiency

• Outperformed state-of-the-art on nuPlan benchmark

Why it matters: Autonomous driving needs robust perception and decision-making in complex environments. Traditional approaches struggle with real-world variability. Hydra-MDP end-to-end architecture overcomes some limitations, potentially leading to safer autonomous driving.

QUICK NEWS

Quick news

• Context caching reduces Gemini API expenses for developers

• 96% of executives urge incorporating AI tools into operations

• Meta FAIR releases new models and research

• Softbank plans to cancel out angry customer voices using AI

TOOLS

🥇 New tools

• Rosie - Your AI phone answering service

• Spinach - Turn meetings into notes

• Sound Effects AI - Create sounds from text or videos