AGENTIC-SYSTEMSOctober 23, 20241 min read14 views

New generative AI tools open the doors of music creation

Google DeepMind has unveiled new generative AI tools, including MusicFX DJ and Music AI Sandbox, signaling a major leap in the intersection of technology and musical creativity.

This work was made possible by core research and engineering efforts from Andrea Agostinelli, Zalán Borsos, George Brower, Antoine Caillon, Cătălina Cangea, Noah Constant, Michael Chang, Chris Deaner, Timo Denk, Chris Donahue, Michael Dooley, Jesse Engel, Christian Frank, Beat Gfeller, Tobenna Peter Igwe, Drew Jaegle, Matej Kastelic, Kazuya Kawakami, Pen Li, Ethan Manilow, Yotam Mann, Colin McArdell, Brian McWilliams, Adam Roberts, Matt Sharifi, Ian Simon, Ondrej Skopek, Marco Tagliasacchi, Cassie Tarakajian, Alex Tudor, Victor Ungureanu, Mauro Verzetti, Damien Vincent, Luyu Wang, Björn Winkler, Yan Wu, and Mauricio Zuluaga.

MusicFX DJ was developed by Antoine Caillon, Noah Constant, Jesse Engel, Alberto Lalama, Hema Manickavasagam, Adam Roberts, Ian Simon, and Cassie Tarakajian in collaboration with our partners from Google Labs including Obed Appiah-Agyeman, Tahj Atkinson, Carlie de Boer, Phillip Campion, Sai Kiran Gorthi, Kelly Lau-Kee, Elias Roman, Noah Semus, Trond Wuellner, Kristin Yim, and Jamie Zyskowski. We give our deepest thanks to Jacob Collier, Ben Bloomberg, and Fran Haincourt for their valuable feedback throughout the development process.

Music AI Sandbox was developed by Andrea Agostinelli, George Brower, Ross Cairns, Michael Chang, Yeawon Choi, Chris Deaner, Jesse Engel, Reed Enger, Beat Gfeller, Tom Hume, Tom Jenkins, Max Edelmann, Drew Jaegle, Jacob Kelly, DY Kim, David Madras, Hema Manickavasagam, Ethan Manilow, Yotam Mann, Colin McArdell, Chris Reardon, Felix Riedel, Adam Roberts, Arathi Sethumadhavan, Eleni Shaw, Sage Stevens, Amy Stuart, Luyu Wang, Pawel Wluka, and Yan Wu in collaboration with our partners in YouTube and Tech & Society.

Dream Track was developed by Andrea Agostinelli, Zalán Borsos, Geoffrey Cideron, Timo Denk, Michael Dooley, Christian Frank, Sertan Girgin, Myriam Hamed Torres, Matej Kastelic, Pen Li, Brian McWilliams, Matt Sharifi, Ondrej Skopek, Marco Tagliasacchi, Mauro Verzetti, Mauricio Zuluaga, in collaboration with our partners in YouTube.

Special thanks to Aäron van den Oord, Tom Hume, Douglas Eck, Eli Collins, Mira Lane, Koray Kavukcuoglu, and Demis Hassabis for their insightful guidance and support throughout the research process. Thanks to Mahyar Bordbar and DY Kim for helping coordinate these efforts, as well as the YouTube Artist Partnerships team for their support partnering with the music industry.

We also acknowledge the many other individuals who contributed across Google DeepMind and Alphabet, including our partners at YouTube.

Source: Google DeepMind Blog

More in this category

agentic-systems

Align AI to Dynamic Human-AI Workflows

Current AI alignment methods fall short by relying on static representations of human preferences. A new paper proposes a paradigm shift toward "interactive and complementary alignment," allowing AI and humans to co-evolve within dynamic workflows.

agentic-systems

MemoHarness: Agent Harnesses That Learn from Experience

Researchers introduce MemoHarness, an adaptive harness optimization framework that enables AI agents to learn from their own executions. By analyzing performance and storing insights in a dual-layer experience bank, it dynamically adapts to new tasks, outperforming static configurations.

agentic-systems

How Artificial Intelligence LLM Engines Shape the Global Conflict Information Environment

A new study reveals how AI-powered answer engines are highly vulnerable to hallucination and manipulation when reporting on global conflicts. The lack of robust digital records allows malicious actors to exploit these systems through Generative Engine Optimization (GEO), posing a new threat of AI-driven information warfare.

agentic-systems

AI Agents Do Not Fail Alone:The Context Fails First

A new study validates context-engineering quality as an independent leading indicator of AI agent reliability, introducing an open-source evaluation infrastructure called ProofAgent-Harness to predict behavioral outcomes before deployment.

agentic-systems

Interpretable Language Model for Closed-Loop Type 1 Diabetes Control

Researchers have developed LLM-T1D, a novel approach combining reinforcement learning and large language models to automate insulin delivery for Type 1 Diabetes. The system not only outperforms traditional methods but also explains its decisions in plain language, addressing the 'black-box' challenge in medical AI.

agentic-systems

RxBrain: Embodied Cognition Foundation Model with Joint Language-Visual Reasoning and Imagination

Researchers have introduced RxBrain, a breakthrough embodied cognition foundation model that unites language reasoning and visual imagination. This model opens new pathways for efficient robot control without requiring massive action-data pretraining.