Hello, I am working on a final project, and a component is voice recognition of keywords. I am able to create and deploy it onto my Arduino BLE 33, but for the project it is required for a user to state a unique keyword for themselves as sort of a password.
We currently have a web server built, and I was wondering if the following is possible:
-User records their keyword on webserver
-Voice Sample dataset sent to Edge Impulse,
-Training and Classification Occurs for the Sample
-Arduino Library returned back to sent back to web server
This is the current approach we are looking into. Is another approach which can allow users to sample their voices and create the classifier without having to go through edge impulse directly.