The OS for ambient devices 🤖 Claim your devkit today 🗣️Join Discord: https://xitter.cf/t.co/nmPij5K9U6

San Francisco, CA
Joined March 2024
500 million smart speakers have been sold. (WILD!) NONE can run AI agents. NONE let you choose your model. NONE are open. BROKEN! We’re building the open alternative FROM SCRATCH! - A new OS - Open source abilities. - Open hardware. Jailbroken. 🤫 Build anything.
46
46
19
596
1,069,338
1/ Picture a kitchen at 7AM. Someone asks the speaker to add milk to the grocery list. Another says "make it two." A kid yells something from the hallway, and the dishwasher starts up. This transcript is a mess. And that's where speaker diarization comes in. A thread 🧵
1
1
2
449
7/ There’s still more to solve. Speaker labels don’t identify people by name, and real-world noise, latency, and far-field audio remain challenges. But diarization models like Nemotron 3 are becoming a building block for voice agents.
1
61
8/ The next generation of voice interfaces won’t just need to understand what was said - but who said it, when they said it, & what else was happening in the room. That’s a key step toward spatial intelligence, and the future of ambient AI. Read more ↓ openhome.com/blog/whos-talki…
57
Voice assistants are getting good at understanding what we say. But what about knowing who said it? NVIDIA recently launched Nemotron 3 Diarization, it's open-weight model for tracking who’s speaking in real time. Here's why we think it matters👇 openhome.com/blog/whos-talki…
1
2
7
327
We’ve spent years making AI models better at chat. But what if the next big leap in voice assistants comes from models that don’t talk at all? Jev, a new kind of AI model built to make fast, structured decisions, might just change the game. A thread 🧵👇
2
1
3
412
8/ A specialized decision model could handle many of these steps quickly, leaving a larger LLM to focus on the reasoning and language that actually require it. That’s a different way to build voice AI: not one model doing everything, but multiple models working together.
1
54
9/ This is especially important as AI moves beyond chat and into our everyday spaces. Ambient AI needs to do more than generate great answers. It needs to understand context, make decisions, act - all at the right moment, without making people wait. openhome.com/blog/voice-assi…
55
ICYMI: Your speaker hears more than you’d ever type into a chatbot. That’s why we’ve integrated @veniceai into OpenHome, bringing private AI reasoning to hardware you own. More control over your AI. More privacy for the conversations happening at home. Read more 👇 openhome.com/blog/venice-ai
1
5
297