ESP32-S3 Home Assistant voice setup with ESPHome

This tutorial converts an ESP32-S3 voice box into a local Home Assistant voice assistant. The presenter first tests its prebuilt DeepSeek setup, including spoken light controls, then walks through Wi-Fi registration and a Home Assistant connection using a HACS integration.

The second route replaces the device firmware with an ESPHome configuration linked in the description. The setup uses Whisper for speech recognition and Piper for speech output through Home Assistant. Viewers follow device creation in ESPHome Builder, copy the encryption key, enter Wi-Fi credentials in the secrets file, and install the compiled firmware over USB through ESP web. The presenter also shows the setting that permits the device to control Home Assistant.

The demonstration covers wake word activation, a single tap to start listening, and a double tap to open status information such as battery level, Wi-Fi connection and wake word status. A wink indicates that the device is ready to listen after boot. The presenter switches the wake word from "Okay, Nabu" to "Hey, Jarvis" and demonstrates light control.

The presenter describes the replacement setup as fully local, but treats it as a personal project rather than a guaranteed conversion. Errors remain visible in the logs, which the tutorial uses for troubleshooting.