SMRTR summary
Google DeepMind's Magenta RealTime 2 music AI now runs live on iPhones — even a 2020 iPhone 12 Pro — generating continuous stereo audio at 25 frames per second without ever using the GPU, by splitting the model into three separate components that each run on the most efficient available hardware.
SMRTR provides this summary for quick context. The original article belongs to Hacker News.
Read the original articleThis archive is built from SMRTR newsletter summaries. Subscribe for hand-picked stories without the extra noise.
More SMRTR summaries that connect to this topic.
Microsoft's TRELLIS.2 image-to-3D model now runs on Apple Silicon Macs without expensive NVIDIA GPUs. Mac users can generate detailed 3D meshes with 400,000+ vertices from images...
Google launched a music generation feature in its Gemini app powered by DeepMind's Lyria 3 model, allowing users to create 30-second songs with lyrics by describing what they want...
Open source DAW plugin using Google's Lyria model to generate live AI music from text prompts in real-time.
ACE-Step generates 4 minutes of music in 20 seconds on 8GB VRAM, solving the speed bottleneck that plagues other AI music models.
Stability AI and Arm are collaborating to bring AI-generated audio to mobile devices with Arm chips. The optimized Stable Audio Open model can generate an 11-second audio sample...
Google's LiteRT framework makes it easier for developers to run powerful AI features on mobile devices by tapping into Neural Processing Units (NPUs) — chips built specifically...