🍖 project · the sensory machine
Meatball can see, hear, and talk back.
A caseless home server built from other people's cast-offs — salvaged GPUs, an abandoned-office pile of audio dongles, a '60s microphone, early-'90s speakers — taught to see, hear, think, and speak. Every model runs on the metal. Nothing touches the cloud. This is its home.
is anyone there?
It can just ask.
Meatball says hi through the Altec Lansings, then listens on the always-on mics. A reply → probably a person. Silence → probably not. It also does this on its own when the cameras catch someone — the simplest sensor there is.
what's moving
It tracks its own motion.
each camera, every ~10s · left = frame + heatmap (where it changed) · right = the raw subtraction · line = motion over time
what's listening
Always-on ears.
no transcriptions yet — speak near a mic and it'll trip the floor and land here
the senses
Four cast-off parts, four working senses.
EarsA '60s garage mic, a thrift-store condenser, and the webcam mics — feeding a local Whisper that transcribes the room.
MouthEarly-'90s Altec Lansings driven by a local neural voice. First word lands in about a third of a second.
BrainA local LLM on two old GPUs, plus a self-calibrating audio rig. No keys, no cloud, ever.
the build
The bill of materials.
salvaged → repurposed
- No case at allopen-air on the bench — a body you can reach into
- Two old GPUsrun the local LLM, Whisper STT, and a neural voice — on the metal
- 20 assorted second-hand external drivesthe memory
- A pile of Plantronics ADACs from an abandoned officethe audio lanes — speaker + mics over USB
- An old factory start/stop buttonthe power switch
- Altec Lansing speakers, early '90sthe mouth
- A mic from my grandfather's garage, '60san ear
- A Realistic condenser mic, Salvation Armyanother ear
- A baseless Salvation Army monitorthe display
- Scrounged Logitech webcamsthe eyes — and, it turned out, the best ears too
field notes