AGENTIC-SYSTEMSApril 24, 20251 min read15 views

Music AI Sandbox, now with new features and broader access

Google DeepMind expands its Music AI Sandbox and introduces Lyria 2, highlighting a massive collaborative effort to transform music creation through artificial intelligence.

Music AI Sandbox was developed by Adam Roberts, Amy Stuart, Ari Troper, Beat Gfeller, Chris Deaner, Chris Reardon, Colin McArdell, DY Kim, Ethan Manilow, Felix Riedel, George Brower, Hema Manickavasagam, Jeff Chang, Jesse Engel, Michael Chang, Moon Park, Pawel Wluka, Reed Enger, Ross Cairns, Sage Stevens, Tom Jenkins, Tom Hume and Yotam Mann. Additional contributions provided by Arathi Sethumadhavan, Brian McWilliams, CÄtÄlina Cangea, Doug Fritz, Drew Jaegle, Eleni Shaw, Jessi Liang, Kazuya Kawakami, Kehang Han, and Veronika Goldberg.

Lyria 2 was developed by Asahi Ushio, Beat Gfeller, Brian McWilliams, Kazuya Kawakami, Keyang Xu, Matej Kastelic, Mauro Verzetti, Myriam Hamed Torres, Ondrej Skopek, Pavel Khrushkov, Pen Li, Tobenna Peter Igwe and Zalan Borsos. Additional contributions provided by Adam Roberts, Andrea Agostinelli, Benigno Uria, Carrie Zhang, Chris Deaner, Colin McArdell, DY Kim, Eleni Shaw, Ethan Manilow, Hongliang Fei, Jason Baldridge, Jesse Engel, Li Li, Luyu Wang, Mauricio Zuluaga, Nemanja Spasojevic, Noah Constant, Ruba Haroun, Tayniat Khan, Volodymyr Mnih, Yan Wu and Zoe Ashwood.

Special thanks to AÃ¤ron van den Oord, Mahyar Bordbar, Douglas Eck, Eli Collins, Mira Lane, Koray Kavukcuoglu and Demis Hassabis for their insightful guidance and support throughout the development process.

We also acknowledge the many other individuals who contributed across Google DeepMind and Alphabet, including our colleagues at YouTube (a particular shout out to the YouTube Artist Partnerships team led by Vivien Lewit for their support partnering with the music industry).

Source: Google DeepMind Blog

More in this category

agentic-systems

Generative Ontology Induction: Domain-Agnostic Schema Discovery from Document Corpora Using Large Language Models

Researchers introduce Generative Ontology Induction (GOI), a domain-agnostic framework that automatically extracts structured ontologies from document corpora using LLMs. Achieving 95-100% structural coverage, GOI addresses a major bottleneck in knowledge-intensive AI systems.

agentic-systems

JUMP: Single-Pass Membership Inference on Fine-Tuned Diffusion Language Models

Researchers have proposed JUMP, a novel single-pass membership inference attack designed for fine-tuned discrete diffusion language models (dLLMs). By leveraging the unique properties of dLLMs, JUMP significantly improves detection accuracy while drastically reducing the number of required queries.

agentic-systems

Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL

Researchers demonstrate that Masked Diffusion Language Models (MDLMs) serve as highly effective, steerable text-based world models for agentic reinforcement learning. By leveraging bidirectional denoising, MDLMs outperform autoregressive models four times their size in coherence, groundedness, and rollout diversity.

NOW LET US Related – Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment

agentic-systems

Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment

A new study demonstrates that small language models (SLMs) under 3 billion parameters can serve as highly capable local experts for specialized tasks. By combining structured benchmarking with low-cost parameter-efficient fine-tuning (PEFT), institutions can achieve AI autonomy without relying on expensive hardware.

agentic-systems

PlanFlip: Attacking Multi-Agent LLM Systems via Planning-Phase Prompt Injection

Researchers have introduced PlanFlip, a novel prompt injection attack framework targeting the planning phase of multi-agent LLM systems. The study reveals critical security blind spots in homogeneous agent pipelines and demonstrates that reasoning-augmented models like DeepSeek-R1 exhibit strong resistance.

agentic-systems

Rater State Bias in RLHF Preference Data: An Audit Framework

A new study identifies a structured bias in Reinforcement Learning from Human Feedback (RLHF) caused by the psychological state of human raters. Under stress, raters' shifting preferences can propagate through reward modeling, potentially compromising AI policy optimization.