- Bot Eat Brain
- Posts
- Udio announced its music generation and sharing app
Udio announced its music generation and sharing app
PLUS: Don’t just analyze, EQA-lize
Michael Parrish
April 16, 2024
Sponsored by
Good morning, human brains, and welcome back to your daily munch of AI news.
Here’s what’s on the menu today:
Don’t waste your life learning music skills 🤤 🎧
Udio launched its music generation app to rival Suno.
Eternal Sunshine of the Spotless Grok 🤖 👀
xAI unveiled Grok-1.5V. It’s the company’s first multimodal AI model.
Don’t just analyze, EQA-lize 🏠 🤔
Meta released OpenEQA, evaluates AI’s understanding of environments.
MAIN COURSE
Your $70,000 music degree is trash 🤤 🎧
On Wednesday, Udio announced its music generation and sharing app. It was developed by former Google DeepMind researchers to rival Suno.
What is it?
If Udio’nt know, you’re missing out… 🤭 It allows you to generate fully produced/mastered tracks by inputting musical genres, personalized lyrics, artist inspirations, and more.
Is it any good?
Early users claim it produces higher quality sounds than competitors like Suno. For science, we fed the same prompt to Suno and Udio:
“A neo-soul song about a robot that eats brains.”
Which song do you think is better? |
What makes this different from other music generators?
It allows you to edit your music’s style, make your tracks shorter or longer, and generate up to 1200 songs a month for free.
I need more music generators in my life.
We got you. Back in December, we reported on Suno. It allows you to create full audio compositions, lyrics, instrumentals, voices, and more using text prompts.
In February, we covered Adobe’s Project Music GenAI Control. It’s an AI model that generates audio from text descriptions or melodies.
A couple of weeks ago, we reported on Stable Audio 2.0. It’s a music and sound effect generation model that can create songs up to 3 minutes long.
SPONSORED BY SIMPLY LAB LIMITED
MaxAI.me -Maximize Efficiency with AI-Powered Browsing
Transform your browsing with MaxAI.me! Instantly summarize articles, videos, and PDFs; and perfect your writing across any webpage with a single click; craft emails, and search the web with AI. Loved by 1M+ users and ranked #1 of the day and the week on ProductHunt, it's your gateway to effortless online efficiency. No more tedious tasks—just seamless, AI-powered magic. Install now and join the revolution of smart browsing!
SIDE SALAD
Once you Grok, you can’t stop 🤖 👀
On Friday, xAI announced Grok 1.5V. It’s xAI’s first multimodal model that excels in multidisciplinary reasoning, spatial relationship analysis, and more.
What in Grok-nation?
Grok 1.5V can analyze visual content like documents, diagrams, and more. It generates code from drawings, creates bedtime stories from children’s art, calculates calories from a photo of food, and more.
Is it Grok-licious or Grok-gone bad?
It absolutely dominates in xAI’s RealWorldQA benchmark. It evaluates AI's understanding of the physical world with over 760 image-based Q&A pairs. How it performs on other benchmarks remains to be seen.
How do I unleash the Grok?
xAI claims Grok-1.5V will be released “soon” to early testers/existing Grok users.
Who started the Grok?
It was always burning. In November, we reported on Grok. It uses real-time world knowledge to wittily answer questions, suggest questions to ask, and more.
In March, we covered xAI’s open-source release of Grok-1. You can download it now; it’s about 300 GB and contains 773 files.
Two weeks ago, we reported on xAI’s release of Grok 1.5. It features improved reasoning capabilities, an extended context length, and more.
A LITTLE SOMETHING EXTRA
OpenEQA-sy does it… 🏠 🤔
On Thursday, Meta unveiled OpenEQA. It’s a benchmark that assesses how AI understands physical spaces.
What is it?
It stands for Open-Vocabulary Embodied Question Answering. It’s a benchmark that evaluates how AI models understand physical environments and answer open-ended questions.
What’s under the hood?
OpenEQA contains over 1,600 question-answer pairs from human annotators drawn from more than 180 videos of physical environments.
What else has Meta been up to?
In December, we covered Meta and IBM’s AI Alliance. The point is to allegedly foster more open AI resources and stable working groups.
The next week, we reported on Meta’s Purple Llama. The purpose is to establish trust in AI development by providing tools for building responsible AI.
A couple of days later, we covered Meta’s Ray Ban smart glasses. It began testing multimodal features that leverage the glasses’ cameras and microphone.
YOUR DAILY MUNCH
Salad
Reduce your cloud expenses by up to 90% with SaladCloud's cost-effective GPU solutions. Start projects at just $0.02/hr. Book a 15-minute call with our experts or get a demo to discover how much you could save. Plus, receive $50 in testing credits! Start saving today with SaladCloud.
Think Pieces
Looking for a job in AI? Here are 74 robotics companies that are currently hiring AI researchers, developers, leaders, and more.
Is AI in healthcare a good or bad thing? Both patients and professionals experience mixed emotions about whether it’s ready for serious use.
What’s up with Meta AI? It’s testing a new chatbot with WhatsApp, Instagram, and Messenger users in India and Africa.
Startup News
OpenAI fired two AI researchers for a suspected data leak. The data is reportedly related to OpenAI’s Project Q, though specifics are still undisclosed.
Humane’s AI Pin received negative reviews. It fails to perform basic functions such as setting alarms, adding tasks to your calendar, and more.
OpenAI claims GPT 4 Turbo is available for all paid users. So far, many users complain about not having access. On our end, we’re still stuck with GPT-4.
Research
Leave No Context Behind — a technique that equips LLMs with “infinitely long inputs.”
RealmDreamer — a 3D image generation technique that achieves state-of-the-art results with parallax, detailed appearance, and high-fidelity geometry.
RULER — a thorough, comprehensive evaluation benchmark for long-context language models.
MEMES FOR DESSERT
TWEET OF THE DAY
Are LLMs like ChatGPT killing the traditional college essay?
Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.
AI ART-SHOW
Until next time 🤖😋🧠