Question/Issue:
I have a proven model in EI website and data sent from esp32s3 through data forwarder. The data is sound recognition. The prediction runs fine, with very good and accurate recognition rate.
Downloading the arduino code and fit to the device, the prediction is not stable, less accurate.
Why? How can I fix it?
The output from the microphone is very same like in data forwarder. Same hardware.
Thanks!