Magenta RealTime 2: An Open, Locally Runnable Real-Time Music Model
Original: Magenta RealTime 2: Open and Local Live Music Models
Google Magenta released open, local live music models for low-latency AI instrument building on Apple Silicon.
Magenta RealTime 2 is an open-weights live music model designed for interactive performance rather than offline prompt-to-song generation. It supports real-time control through MIDI, audio, and text, and can run as standalone apps, DAW plugins, or embedded music software. Google Magenta also released a Python library, C++ MLX inference engine, models, and example applications for musicians and developers.
Google Magenta has released Magenta RealTime 2 (MRT2), positioning it as an open-weight, locally runnable real-time music model. Its focus is not on generating a complete track from a prompt in one pass, but on allowing the model to be played and controlled in real time like an instrument. MRT2 supports low-latency control through MIDI, audio, and text inputs, and can be used as a standalone application, a DAW plugin, or integrated into other music software. The official release includes a 2.4B-parameter Base model, a 230M Small model, a Python library installable via pip install magenta-rt, inference support based on JAX / MLX / SequenceLayers, and a streaming audio generation engine written in C++ for Apple Silicon GPUs.
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on Hacker News (AI keywords) →Related
Summaries are AI-generated; the original article is authoritative.