esp32-tensorflow-microspeech

ESP32 wake word detection with tensor flow. This project demonstrates how signals from a I2S microphone can be processed to detect keywords such as "yes" and "no".

Hardware

This project is running on the LillyGo TTGO T-Camera and nearly reaches 5 inferences per second.

Running the code

Download/ clone this repository and open it with Platformio in VSCode. Compile it and flash it to your ESP32. The board requires a microphone to properly work.

Credits

This sample was created based on the TinyML book by Pete Warden, Daniel Situnayake. Getting the code to run in platformio for ESP32 and the Arduino platform was done with the help of Wezley Sherman's Medium Blog Post