MusicLM is wild. Very realistic and versatile, can condition on text, images, and other audio. Coolest feature: hum or whistle a melody, input some text (e.g. "string quartet") and it spits out a string quartet playing that melody! https://
google-research.github.io/seanet/musiclm
/examples/
…
RESEARCH
-
MusicLM: Google’s AI Generates Music From Text and Audio
By
–
-
David Wajc Wins SODA 2023 Best Paper Award
By
–
Congrats to visiting faculty David Wajc and co-authors of "Dynamic Matching with Better-than-2 Approximation in Polylogarithmic Update Time" for winning the SODA 2023 best paper award (
https://
goo.gle/3iZKbMp). Learn more about the work ↓
https://
goo.gle/3R52goS -
Dataset Preparation for AI Training in Two Lines Code
By
–
With just 2 lines of code the dataset would be ready for your training workflows!
-
Train Your Own MusicLM with MusicCaps Dataset
By
–
Want to train your own MusicLM? 🎶
— Vaibhav (VB) Srivastav (@reach_vb) 27 janvier 2023
The MusicCaps dataset is now on the 🤗Hub: https://t.co/JKrVYng9OY
The MusicCaps dataset contains 5,521 music examples, each of which is labeled with an English aspect list and a free text caption written by musicians. 🎸 pic.twitter.com/Byiy9r7hzsWant to train your own MusicLM? The MusicCaps dataset is now on the Hub: https://
huggingface.co/datasets/googl
e/MusicCaps
… The MusicCaps dataset contains 5,521 music examples, each of which is labeled with an English aspect list and a free text caption written by musicians. -
AI Investment Predictions for 2023 and Beyond
By
–
What kinds of investments can we anticipate in the #AI space in 2023 and beyond? See what DataRobot CTO, Michael Schmidt, predicts for the future of AI on @wef . Read more
-
AI Progress Unpredictability and Human Capability Assessment
By
–
One lesson from AI progress is it’s hard to predict the relative difficulties of various skills; one lesson from AI limitations is that humans are much more capable than we give ourselves credit for.
-
Meta Myths: Point by Point Affirmation Analysis
By
–
Sounds like a point by point affirmation of Meta Myths
-
Google LolNerf: Learn from One Look at Images
By
–
Learn more about it here – https://
ai.googleblog.com/2022/09/lolner
f-learn-from-one-look.html
… -
LOLNeRF: Combining GLO and NeRF for Novel View Synthesis
By
–
How does it work? The combination of Generative Latent Optimization (GLO) and neural radiance fields (NeRF) used in LOLNeRF is what sets it apart from other approaches. It achieves SOTA results for novel view synthesis and competitive results for depth estimation.
-
LOLNerf: Learning 3D Object Recognition From Single Images
By
–
LOLNerf: Learn from One Look”, we explore the ability to learn a high-quality representation from just a single 2-D image. It can understand common 3D objects, like cars and human faces, from just one image – a game changer for object recognition.