AxiomVox: Portable Voice Capture and Transcription

AxiomVox is my portable AI voice capture and transcription project. The idea is to build a small appliance-style device that can boot reliably, show useful local status, capture audio, and eventually hand that audio into a broader transcription workflow without needing to feel like a fragile science project every time it powers on.

The current target is Raspberry Pi hardware, including the Raspberry Pi Zero W and Zero 2 W, with Raspberry Pi OS Lite, a PiSugar Whisplay HAT, PiSugar 3 UPS, and HDMI as a passive secondary status display. The device side is written in Python and managed by systemd, with a local web dashboard for status, controls, logs, and graceful shutdown. The repo also has a future-facing server area and shared protocol/state code so the project can grow beyond a single box.

Most of the recent work has been about making it behave more like an appliance than a loose pile of scripts. The milestone notes walk through that progression: hardware validation and boot status, the core appliance loop, local recording-session metadata, WAV capture and validation, settings and power controls, searchable logs, sound feedback, and CPU/RAM graphs. It is the kind of project where the boring parts matter because the goal is trust: when the button gets pressed, the device should know what it is doing and tell me what is happening.

The GitHub repo is here: github.com/drgncabe/AxiomVox. It is still evolving, but the direction is clear enough now: portable voice capture first, transcription pipeline next, with enough local feedback and diagnostics that it can be used away from a desk without guessing whether it is alive.