This project involves developing a compact, low-power voice assistant using a single ESP32-S3 microcontroller, a microphone, and a speaker. The assistant answers queries by processing voice input through Google’s Gemini AI and communicates responses via Text-to-Speech. It can handle general questions, math problems, and real-time translations. Designed for portability and affordability, it offers a cost-effective alternative to setups using Raspberry Pi or cloud services.
The system functions by recording voice input, transcribing it with the Deepgram API, retrieving a response from the Gemini AI API, and playing it back through a speaker. Key components include the ESP32 WROOM32D, MAX98357 amplifier, INMP441 MEMS microphone, and a LiPo battery for power.
This project highlights the potential of AI in embedded systems and aims to inspire educators and developers to explore AI and IoT applications, making advanced technology accessible for a variety of uses, from education to prototyping.
Source link
