v1.0.0

Waveshare ESP32-S3 Voice Assistant — v1.0.0

First tagged release. Firmware for a touchscreen ESP32-S3 device that runs
a Gemini Live voice assistant and a networked music player, replicating its
state to a companion starhub daemon over the STAR WebSocket protocol.

Highlights

• Gemini Live voice assistant — direct WebSocket connection to Google's
  Gemini Live API with a full tool-calling surface (playback, volume, LED,
  SD-card file I/O, alarms, memory).
• Music playback — Invidious-backed streaming via media_player, SD
  card caching, catalog DB, WebM/Opus decoding.
• On-device touch UI — LVGL v9 dashboard on the ILI9341 + XPT2046
  display (ui_view), gated behind CONFIG_DISPLAY_ENABLE.
• STAR replication — core_sysdb's WAL-based state replication keeps
  the device and the starhub daemon in sync over a single WebSocket
  command channel.
• Split into three repos: waveshare (firmware, this repo),
  starhub (replication daemon),
  and waveshare-dashboard
  (React control UI) — each with preserved git history, now deployable
  together via Docker Compose.
• CI — GitHub Actions now builds the firmware (ESP-IDF v6.0.1,
  esp32s3) on every push/PR, catching build-reproducibility regressions
  that a gitignored local sdkconfig was previously masking.
• Docs — fact-checked, corrected README plus a README.md for every
  major component (core_sysdb, audio_core, media_player,
  gemini_live, ui_view).

Known limitations

• MQTT (mqtt_forward tool, Mqtt SysDb component) is defined but not
  implemented — no live client exists.
• MusicLibraryManager is dead code, superseded by CatalogDB.