myAIpal
Posts
Major Labels Sue AI Music Generators

Major Labels Sue AI Music Generators

Plus: How to build a Siri-like voice assistant

Tosin Paul
July 18, 2024

Hello! AI Fanatics

Last week, China released a Sora-like AI video model called KLING, and now the US is back in the race. Luma AI has introduced Dream Machine, a video AI model that is already showing a lot of promise and is available for use. Additionally, Amazon has pledged $230 million to support generative AI startups, fueling innovation and pushing the boundaries of what's possible. In other news, Harvard and Google's DeepMind have unveiled a virtual rat with an AI brain, designed to replicate the neural activity and movements of real rats. Let’s dive in!

News Snippets

^OPENAI

👨🏻‍⚖️ Suno and Udio Sued by Major Record Labels

Image source: Midjourney

Universal Music Group, Warner Music Group, and Sony Music Entertainment just filed lawsuits against Udio and Suno, alleging the music generation startups illegally trained their AI models on copyrighted material.

THE BREAK DOWN:

The Recording Industry Association of America is spearheading the lawsuits, seeking up to $150,000 per infringed work, with potentially billions in damages.
The labels claim Suno and Udio trained their AI models on copyrighted recordings without permission or compensation.
The RIAA alleges that the platform's outputs mimic specific artists and hit songs and that the companies have admitted to training on copyrighted material.
Both startups have raised significant funding recently, with Suno securing $125M in May and Udio receiving $10M.

The same legal issues we saw face text and image generation platforms have come for AI music — but unlike the other sectors, record labels have seemed more proactive in looking to work with AI firms. Time will tell if these latest suits will set actual precedents for the industry, or just hefty settlements and eventual licensing deals.

^SYNTHESIA

🗣️ Synthesia’s major AI avatar upgrades

Image source: Synthesia

AI avatar startup Synthesia just unveiled a major platform upgrade, introducing a series of new features, including full-body AI avatars, AI Video Assistant, translation, and more.

THE BREAK DOWN:

New ‘Personal AI Avatars’ allow users to create digital twins of themselves using HD cameras, a webcam, or a phone, with lip-syncing and realism.
Synthesia also teased full-body avatars coming ‘later this year’, which will be fully controllable and have more personality and animation options.
An upgraded AI Video Assistant will be capable of producing entire video libraries from existing knowledge bases, complete with brand-specific elements.

Why it matters: Just last month, the founder of Zoom said a world is coming where our digital twins attend meetings for us — and while we aren’t there yet, AI avatar platforms are getting into territory that’s nearly indistinguishable from reality. Video communication is about to be redefined sooner than most people think.

Trends & Predictions:

^AMAZON

🤖 Amazon is secretly working on a ChatGPT killer

Image source: Amazon

Amazon is developing Metis, an AI chatbot to rival ChatGPT.

THE BREAK DOWN:

Metis will provide text and image-based responses, offer source links, and suggest follow-up questions.
Built on Amazon's Olympus model, Metis aims to outperform competitors like ChatGPT and Claude.
The AI will automate complex tasks and integrate closely with theAlexa project.

Amazon is developing an AI chatbot named Metis, designed to compete directly with ChatGPT. Accessible through a web browser, Metis will deliver both text and image-based responses, featuring source linking, follow-up question suggestions, and image generation. The AI will leverage Retrieval Augmented Generation to access current information, enhancing its capability to provide real-time data like stock prices. Metis will also act as an AI agent, capable of automating complex tasks and closely integrating with the "Remarkable Alexa" project. It is rumored that Amazon will launch "Remarkable Alexa" with a new subscription model. Powered by Amazon's Olympus model, Metis boasts hundreds of billions of parameters and is expected to surpass competitors like ChatGPT and Claude. Amazon plans to launch Metis in September, coinciding with a significant Alexa event.

Ai Training:

AI TRAINING

🗣️ Build a Siri-like voice assistant

Image Source: Siri

This custom iPhone shortcut lets you create a lightning-fast, Siri-like voice assistant powered by the Llama 3 language model and Groq's free API keys.

Step-by-Step:

Sign up for a free account at Groq and generate an API key.
Create a new shortcut in the iPhone Shortcuts app.
Add actions to capture voice input, store your API key, and set the Groq API endpoint for Llama 3.
Call the Llama 3 API using a "Get Contents of URL" action with the appropriate headers and request body.
Extract the generated response and use the "Show Results" action to have your iPhone show it to you!