
We had so much fun building MRT2. Proud to see it go live. Here's a chord steering example. Name that tune anyone?

Research Scientist @GoogleDeepMind

We had so much fun building MRT2. Proud to see it go live. Here's a chord steering example. Name that tune anyone?

Introducing Magenta RealTime 2 (MRT2): the live music model you can play as an instrument. MRT2 offers MIDI and prompt controls, and runs natively on a MacBook with <200ms latency. Open weights. Open source inference engine. Suite of apps and plugins. Hear what it can do and try it out for yourself below 🧵

Today, we're open sourcing the code behind "The Infinite Crate," our VST/AU plugin that lets you play with Lyria RealTime directly inside your favorite DAW! 🎧🎶 Since we released The Infinite Crate last year, it’s been used by some of our favorite artists - including a wonderful showcase with @daitomanabe in Tokyo - and being featured as a top new music tool at NAMM 2026. We’re now fully open sourcing the plugin for developers to fork, modify, and make their own under the permissive Apache 2.0 license. 💾 Get it here: github.com/magenta/the-infin…

Magenta is hiring student researchers for 2026! We're looking for candidates to push the boundaries of interactive real-time music generation, such as the next generation of Lyria RealTime and Magenta RealTime models. Follow instructions to apply here: forms.gle/EAvTt1KLNqpVHx9h7

We built a web app that lets you fly a spaceship through a 3D constellation of music - powered by our Lyria RealTime model. 🎶 Space DJ is an interactive visualization where every star represents a different music genre. As you explore, your path is translated into prompts for the API, creating a continuously evolving soundtrack. ↓

We’re just getting started with live music models. Can’t wait to see the possibilities for musicians of any skill level to craft, shape, layer their sound in real time.

Would love to see this combined with fine-tuned models for more personalized results. colab.research.google.com/gi…

Or sometimes it’s like a looper pedal with a mind of its own? Hard to describe, so try it out yourself! colab.research.google.com/gi…

Audio Injection lets you steer Magenta RT using *live* audio input. Doesn’t feel like jamming with a human. More like jamming along with the radio... but then the radio starts incorporating elements of your sound??
Realtime interactive generative models FTW! Announcing a new 🌊 of details and features for Magenta RealTime, the open weights live music AI model from GDM! * Live Jamming with audio input 🎤🎸🎵 * Personalize your own models 🔧 * Tech report 📜 Links below in the 🧵...

Made a walkthrough vid for Magenta RealTime “Audio Injection”! The notebook takes ~10m to spin up, but totally worth it for the surreal experience 🎤💻🎧⁉️

Everyone talks about real-time AI for speech but sleeps on it for music Real-time is nice to have for conversations, but essential for co-jamming
Realtime interactive generative models FTW! Announcing a new 🌊 of details and features for Magenta RealTime, the open weights live music AI model from GDM! * Live Jamming with audio input 🎤🎸🎵 * Personalize your own models 🔧 * Tech report 📜 Links below in the 🧵...

Hours and hours of fun and inspiration. Tiny excerpt: youtube.com/shorts/53gmT8zaf…

🔊 Easy to lose hours to live jamming w/ Magenta RT. Feels like a glimpse at something magical Try it yourself! Some setup involved (we're working on improving that!) but worth it: g.co/magenta/rt-audio-inject…

In 2018, we wrote a paper titled "the challenge of realistic music generation". A few years ago, this challenge was essentially solved: it is now a thing that exists🤯 Realtime is the next frontier, it opens up so many more possibilities. There's a tech report + open weights!🙏
Realtime interactive generative models FTW! Announcing a new 🌊 of details and features for Magenta RealTime, the open weights live music AI model from GDM! * Live Jamming with audio input 🎤🎸🎵 * Personalize your own models 🔧 * Tech report 📜 Links below in the 🧵...

I’m happy to have donated my voice to this incredible new open model and instrument! It is a new, active, approach to spawning music, made for real time jamming and exploration. More like a sampler than a jukebox, you can fine tune it on your own sounds and soon run it locally!
We've worked with @hollyherndon and @matdryhurst to finetune Magenta RT on Holly's vocals and the result is a really cool model that keeps the base model's ability to respond to wide-ranging controls while infusing outputs with the unique sonic qualities of the finetuning domain

We've worked with @hollyherndon and @matdryhurst to finetune Magenta RT on Holly's vocals and the result is a really cool model that keeps the base model's ability to respond to wide-ranging controls while infusing outputs with the unique sonic qualities of the finetuning domain
We're thrilled to have worked with Holly Herndon to train a model finetuned on her vocal style to share with all of you. Playing with Holly’s model, you can use additional controls to get vocalizations that contour the music or "color" the instruments with a timbre that resembles Holly's voice