Natural Language Processing with Tiny ML Dataset

Natural Language Processing Dataset for research into a complex keyword model. Data is split into two models one for colors and another for commands. These keywords were all trained on one voice on the Arduino Nano 33 BLE sense chip's microphone. The data itself exists in both .wav and .json formats and can be imported to Edge Impulse for proper use or the .wav files can be used for model training outside of Edge Impulse formats. The dataset itself also contains a test and train split and each file has an identifier for its specific label.