AI Newsletter #91 - Google generating realistic and consistent audio

Welcome to Nural's newsletter focusing on how AI is being used to tackle global grand challenges.

Packed inside we have

  • Tesla CEO Elon Musk unveils prototype humanoid Optimus robot
  • Meta and Google AI systems that generate videos from text
  • and AudioLM: a Language Modeling Approach to Audio Generation

If you would like to support our continued work from £1 then click here!

Marcel Hedman


Key Recent Developments


Tesla CEO Elon Musk unveils prototype humanoid Optimus robot

Article


Generative AI: A Creative New World

Generative AI: A Creative New World
Sequoia Capital

What: A comprehensive overview of the generative AI landscape at the application layer. Beyond the most well known text and image generation applications, we are seeing the emergence of code, speech and 3D image generation.


Facebook Research - Introducing Make-A-Video: An AI system that generates videos from text

Introducing Make-A-Video: An AI system that generates videos from text
Make-A-Video builds on Meta AI’s recent research in generative technology and has the potential to open new opportunities for creators and artists.
Google’s Imagen takes on Meta’s Make-A-Video as text-to-video AI models ramp up
Thanks to Google and Meta, the text-to-video trend shows all the signs of getting ready to explode much like text-to-image tools.

What: Meta AI have released a model which turns text into short form videos, building upon recent successes by many models to generate images from text prompts. Th opportunities this unlocks for content creation is astounding. Google wasted no time in responding and have released their own text to video model. Generative AI is moving at a rapid pace, will safeguarding be able to keep up?


AudioLM: a Language Modeling Approach to Audio Generation

AudioLM: a Language Modeling Approach to Audio Generation
Posted by Zalán Borsos, Research Software Engineer, and Neil Zeghidour, Research Scientist, Google Research Generating realistic audio re...

What: Google AI have released a model that can generate realistic and consistent audio given a few seconds of either speech or piano playing. This represents a significant milestone given the complexity of generating both realistic audio alongside audio that is stylistically consistent with initial audio inputs.

The team also trained a classifier that can detect between synthetic audio vs real audio which is a great demonstration of responsible AI.


AI Ethics

🚀 The Biden Administration Proposes ‘A.I. Bill of Rights’

🚀 US regulators call the Stable Diffusion Model by Stability AI unsafe

🚀 Bruce Willis denies selling the rights to his face despite appearing in deepfake Russian commercial

🚀 EU: New liability rules on products and AI to protect consumers and foster innovation

🚀 Leading lawmakers pitch extending scope of AI rulebook to the metaverse

Other interesting reads

🚀 Expert-level detection of pathologies from unannotated chest X-ray images via self-supervised learning [Nature paper]

🚀 DALL·E Now Available Without Waitlist

🚀 NLP consultant must knows

🚀 DeepMind advances AI safety with new Sparrow chatbot


Cool companies found this week

Climate

Everest Labs - An AI-Enabled Operating System for Recycling. They have recently raised $12m to continue their journey of using AI to detect waste from recycling.


...and Finally

Despite the company claiming above that they have authorisation to create a deepfake, Bruce Willis has now claimed that this authorisation was never provided.

AI/ML must knows

Foundation Models - any model trained on broad data at scale that can be fine-tuned to a wide range of downstream tasks. Examples include BERT and GPT-3. (See also Transfer Learning)
Few shot learning - Supervised learning using only a small dataset to master the task.
Transfer Learning - Reusing parts or all of a model designed for one task on a new task with the aim of reducing training time and improving performance.
Generative adversarial network - Generative models that create new data instances that resemble your training data. They can be used to generate fake images.
Deep Learning - Deep learning is a form of machine learning based on artificial neural networks.

Best,

Marcel Hedman
Nural Research Founder
www.nural.cc

If this has been interesting, share it with a friend who will find it equally valuable. If you are not already a subscriber, then subscribe here.

If you are enjoying this content and would like to support the work financially then you can amend your plan here from £1/month!