StudioReddit user built a home Star Trek-style computer, Ada, plus diagnostic panel BM-01, and shares a full postmortem
Original title: I wanted the Star Trek computer in my house, so I built one — meet Ada and BM-01
A user built Ada, an always-on conversational home AI, and just finished BM-01, an ESP32-based physical telemetry/diagnostic panel. The system runs on a Lenovo ThinkCentre M720q with GPT-Live realtime voice, a ReSpeaker XVF3800 four-mic array, local wake-word detection, persistent household memory, WiZ lighting and Samsung TV control, and a Red Alert emergency state. The author's main lesson is that the AI model isn't the hard part; infrastructure like the audio chain, barge-in handling, and memory relevance is. The writeup details migrating to a WebRTC/GPT-Live conversational core, a memory system that evolved into semantic embeddings with categorization, and testing a Microsoft custom keyword model in shadow mode against the existing ONNX wake detector.
The author's months-long firsthand postmortem of a resident home voice AI offers directly reusable lessons on the audio chain, wake words, and memory systems.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.