I want to turn an ESP32 and a Raspberry Pi into a hands-free ChatGPT companion. The ESP32’s only job is local wake-word detection and any other ultra-light edge processing we decide is useful; once the keyword is heard, raw audio from an I2S microphone should stream to the Pi. On the Pi, please capture the audio, run it through an ASR engine powered by my preferred choice—OenAI—and forward the transcribed text to the ChatGPT API. The reply then needs to be converted to speech through OpenTTS and played back on a connected speaker with minimal latency. Alongside the working prototype, I need: • well-commented Python (or C++ if you have a strong rationale) for both boards • wiring diagrams and parts list specific to the I2S mic and any amplifier you select • a simple installer or script that sets up dependencies on a clean Pi OS image • instructions for flashing and configuring the ESP32, including how to update the wake-word model If you can demonstrate sub-second round-trip times and reliable wake-word accuracy, that will be my acceptance benchmark.