[soulofmischief]

I'm tinkering with the idea of an IoT device which sits at the top of your IoT chain. It has microphone and other potential sensory input, which it processes completely offline and turns into text commands which get sent to other devices over the network.

I don't want 10 IoT devices each with their own mics and cameras. I want one device from a brand I trust, which veritably doesn't phone home other than updates, with a dirt-simple API for integration. All other devices can take commands from this one. The age of Alexa and Google Home and "smart TVs" with microphones needs to disappear.

If tenable, the idea could be extended to other sensitive domains such as payment processing with a token system.

Just a thought. Could be hackathon material if I find a good partner.

[komali2]

Just a thought, my old company ElectricImp could be a good fit. The company itself only provides diagrams for compatible hardware, an OS, and a cloud for deployment and a paired cloud-based VM for each deployed.

The only phoning home is OS updates, and then whatever you program the VM to do. So you could take audio in on the device, send to cloud VM, then send to some sort of homebrew speech to text (or be a total fucking badass and squeeze your own speech to text in squirrel code onto the VM itself).