WaivOps HH-TRP Dataset For Music ML

HH-TRP is an open audio dataset of 15,000 drum loops in the style of modern urban trap music, provided in uncompressed stereo WAV format with paired JSON labels. Generated using an algorithmic framework with MIDI patterns and one-shot samples, it includes data augmentation through random sample-swapping to create varied drum kits. The dataset is designed for training and fine-tuning AI models to learn high-performance drum notations for prompt-driven generation and other supervised learning tasks in music applications such as text-to-audio, tempo detection, rhythm analysis and sound design.