From fecbd2c066ca221edafe3b668b70ff66bad89c71 Mon Sep 17 00:00:00 2001 From: eddy Date: Sun, 13 Sep 2026 23:35:27 +0200 Subject: [PATCH] ATOM VoiceS3R HA voice satellite guide --- README.md | 23 +++++++++++++++++++++-- 1 file changed, 21 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index b24d118..274ddd4 100644 --- a/README.md +++ b/README.md @@ -1,3 +1,22 @@ -# atom-voices3r-ha-guide +# ATOM VoiceS3R as Home Assistant Voice Satellite -M5Stack ATOM VoiceS3R as Home Assistant voice satellite: ESPHome config, Wyoming voice pipeline, automations \ No newline at end of file +Turn an M5Stack ATOM VoiceS3R (ESP32-S3 + 8 MB PSRAM) into a compact, +fully offline Home Assistant voice satellite. + +What is in here: + +- `esphome/atom-echos3r-voice-assistant.yaml` — ESPHome config importing + M5Stack's official `atom-echos3r-satellite-base` package (I2S audio, + `voice_assistant`, on-device wake word). Fill in your own keys. +- `docs/build-plan.md` — full step-by-step build plan (Dutch): Wyoming + voice pipeline (Whisper + Piper), flashing, pairing, troubleshooting. +- `automations/s3r-rooktest.yaml` — first spoken-notification automation + as smoke test. + +Hardware: ESP32-S3-PICO-1-N8R8, ES8311 codec, MEMS mic, NS4150B + 1 W +speaker, IR transmitter. The 8 MB PSRAM enables on-device wake word +(microWakeWord, "Okay Nabu") — no server needed for hotword detection. + +Sources: [M5Stack docs](https://docs.m5stack.com/en/core/Atom_EchoS3R), +[M5Stack esphome-yaml](https://github.com/m5stack/esphome-yaml), +[HA Assist](https://www.home-assistant.io/voice_control/).