Over time, I've found different uses for it, and the most recent way that I'm using it is likely the one I'll settle on.
Authored by embedded ML specialists with extensive experience in ESP32 voice recognition architecture, TinyML optimisation, ...
esp32:esp32:esp32:UploadSpeed=921600,CPUFreq=240,FlashFreq=80,FlashMode=qio,FlashSize=4M,PartitionScheme=huge_app,DebugLevel=verbose,PSRAM=disabled,LoopCore=0 ...
I used real board, selected&connected to PC (ESP32-S3-ETH) and ESP32S3 Dev Module, just selected, but not connected. Windows 11, Arduino IDE 2.3.6, ESP32 Core 3.3.2. Open Blank sketch, select ESP32S3 ...