I’ve created a repo to show people how to collect data, train a model on Edge Impulse, and use the continuous audio (keyword spotting) library on STM32CubeIDE: https://github.com/ShawnHymel/ei-keyword-spotting. I plan to go through it in my Hackaday Remoticon workshop, but feel free to work through it if you want to run keyword spotting on an STM32. Feedback is definitely welcome!
This is extremely cool, thank you for sharing! I love your dataset_curation.py script.
@ShawnHymel awesome work as always!