AI Music Editor Transforms Songs with Text Prompts
A new AI system, Instruct-MusicGen, allows users to edit existing music using text instructions. Developed by Queen Mary University of London, Sony AI, and MBZUAI’s Music X Lab, it enhances Meta’s MusicGen model.
Key features:
- Adds, removes, or separates instruments via text prompts
- Requires minimal extra computing power (8% more parameters, 5,000 additional training steps)
- Excels at common music production tasks
The researchers have open-sourced the code, weights, and examples to encourage further development.
In related news, YouTube Music is rolling out a sound search feature and testing AI-powered “conversational radio.”
Microsoft’s AI Designer App Goes Mobile
Microsoft’s AI-powered Designer is now out of preview and available worldwide, bringing advanced image generation, editing, and design capabilities to users on mobile and Windows platforms.
Key details:
- Available in over 80 languages on web, Android, iOS, and Windows
- Uses AI to generate images and designs from text prompts
- Creates outputs like custom stickers, emojis, avatars, and more
- Free version offers 15 daily ‘boosts’ for AI creations (100 for Copilot Pro subscribers at $20/mo)
New features:
- ‘Prompt templates’ for fast creation
- ‘Restyle’ for remixing existing images
- ‘Frame’ for creating personalized frames and collages
This move positions Microsoft in the competitive AI design space alongside Canva and Adobe, as the AI boom reshapes how people approach design.
Anthropic and Menlo Ventures Launch $100M AI Startup Fund
Anthropic is partnering with Menlo Ventures to create the $100 million Anthology Fund, aimed at supporting early-stage AI startups and promoting Anthropic’s technology adoption.
Key points:
- Inspired by the 2008 Apple-Kleiner Perkins iFund partnership
- Menlo Ventures provides investment capital
- Anthropic offers $25,000 in credits for startups to use its large language models
- AI innovation moving “10 to 100 times faster” than previous tech waves
Strategic implications:
- Positions Anthropic competitively against OpenAI’s $175M startup fund
- Aims to foster next-generation AI companies on Anthropic’s infrastructure
- Could strengthen Anthropic’s market position in the long run
This collaboration blends financial and technological support to keep pace with the rapidly accelerating AI development landscape.
Spotify Expands AI DJ to Spanish-Speaking Markets
Spotify is rolling out its AI DJ feature in Spanish, bringing personalized music and commentary to millions of Spanish-speaking Premium users in select markets.
Key details:
- AI DJ combines personalized playlists with AI-generated commentary
- Olivia “Livi” Quiroz Roa, a Senior Music Editor in Mexico City, is the Spanish voice model
- Users can switch between English and Spanish versions in the app
- Launching in Spain and 17 Latin American countries, including Mexico, Argentina, and Colombia
Implications:
- Taps into the massive influence of Spanish language and Latin music globally
- Aims to boost engagement and retention in a huge market
- Spotify reports DJ listeners spend more time on the app and discover more music
This expansion caters to Spanish-speaking users and Latin music fans, offering a personalized AI DJ experience in their preferred language. It represents Spotify’s strategic move to enhance user experience and market presence in Spanish-speaking regions.
OpenAI Launches Cost-Efficient GPT-4o Mini Model
OpenAI has unveiled GPT-4o mini, a compact and affordable version of its flagship GPT-4o model, aimed at expanding AI accessibility for developers and businesses.
Key features:
- Priced at 15 cents per million input tokens and 60 cents per million output tokens
- Over 60% cheaper than GPT-3.5 Turbo
- Scores 82% on the MMLU benchmark, outperforming Google’s Gemini Flash (77.9%) and Anthropic’s Claude Haiku (73.8%)
- Supports a 128K token context window
- Handles text and vision inputs, with audio and video capabilities planned
Implications:
- Replacing GPT-3.5 Turbo in ChatGPT for Free, Plus, and Team users
- Lowers the barrier to entry for AI integrations
- Marks a significant improvement over GPT 3.5 Turbo
This release represents a major step in making advanced AI models more accessible and affordable, potentially accelerating AI adoption across various sectors.
Add AI Sound Effects to Videos with Eleven Labs
Key steps:
- Sign up at Eleven Labs to get credits
- Select ‘Sound Effects’ from the left panel
- Enter a prompt describing desired sound effect
- Click ‘Generate Sound Effect’ button
- Choose from 4 different sound effects generated
- Download and add to your videos
This tool allows users to create custom AI-generated sound effects for video content, enhancing production value with personalized audio elements.
ChatGPT-maker OpenAI Explores Developing Its Own AI Chips
OpenAI, creator of ChatGPT, is in early talks with semiconductor designers, including Broadcom, to develop custom AI chips. This move is part of a larger strategy to expand computing capacity and reduce reliance on Nvidia’s GPUs.
OpenAI CEO Sam Altman has ambitious plans to raise up to $7 trillion for a massive chip-making project. The company is hiring former Google employees with experience in tensor processing units, aiming to overcome the shortage of expensive GPUs crucial for AI model development.
This initiative aligns with an industry trend, as Microsoft, Google, and Meta are also developing in-house chip solutions. Altman is engaging with chipmakers, Microsoft, government bodies, and financial backers. He has met with UAE’s top security official and US Commerce Secretary to discuss the project.
In a related development, the US government announced a $5 billion investment in the chip industry, prompting major chipmakers to invest in the US. This investment is helping companies to kickstart their in-house chip ambitions.
OpenAI’s move reflects the growing demand for specialized AI hardware and its desire to control its technological infrastructure in the rapidly evolving AI landscape.
Mistral AI and NVIDIA Unveil Mistral NeMo 12B, a Cutting-Edge Enterprise AI Model
Mistral AI and NVIDIA have released Mistral NeMo 12B, a state-of-the-art language model for enterprise applications. This model supports chatbots, multilingual tasks, coding, and summarization.
Key features:
- 128K context length for processing complex information
- Released under Apache 2.0 license
- Uses FP8 data format for efficient inference
- Packaged as an NVIDIA NIM inference microservice
The model was developed using:
- NVIDIA DGX Cloud AI platform
- NVIDIA TensorRT-LLM for accelerated inference
- NVIDIA NeMo development platform
Mistral NeMo excels in multi-turn conversations, math, common sense reasoning, world knowledge, and coding. It’s designed to fit on a single NVIDIA L40S, GeForce RTX 4090, or RTX 4500 GPU.
The model was trained using Megatron-LM with 3,072 H100 80GB Tensor Core GPUs on DGX Cloud. It can be deployed in cloud, data center, or RTX workstation environments.
Mistral NeMo is available as an NVIDIA NIM via ai.nvidia.com, with a downloadable version coming soon. This collaboration showcases NVIDIA’s support for the model-builder ecosystem and Mistral AI’s expertise in training data.
This Week’s Top AI Tools
- Liminal: Workflow productivity tools with secure access to gen AI anywhere. Enterprise-grade data security & governance.
- Claude for Android: Anthropic’s Claude is now available as an Android app.
- Morphic Studio: AI-powered platform to create controlled videos.
- Buildbox: Create, design, and build games with AI by just typing.
- Ssemble: Reassemble long videos into engaging shorts.
- Sketch2scheme: Turn diagram sketches into digital schemes.
- Blaze: AI-powered tool for targeting high-intent leads on social platforms.
- Jobright AI: Matches job seekers with tailored opportunities using AI.
- AutoReels.ai: Generates and automates faceless videos for TikTok and YouTube.
- Project Atlas Desktop: Creates business automation agents using natural language.
- Kompas AI: Offers in-depth, accurate searches for professionals with multi-agent verification.
- Fonts Ninja: AI-powered font discovery and organization tool.
- Prodia: Add AI image generation to your app with one API.
- CharacterGen: Efficient 3D character generation from single images.
- Undermind: AI assistant for thorough literature research.
- Shadow: Automates meeting follow-up tasks using AI.
- Superjoin: Imports live data into Google Sheets automatically using AI.
- Almeta ML: Predicts customer behavior to increase revenue with machine learning.
- Microsoft Designer Mobile App: Smartphone designer tool powered by AI.
What’s your take on these rapid AI developments across industries? Are you already using AI tools in your work? We’d love to hear about your experiences with AI-powered music editing, design tools, or enterprise solutions. Share your thoughts on which industry you think will be most dramatically transformed by AI in the coming years!
