The Upright Piano Dataset is a comprehensive repository of audio tracks paired with detailed metadata, carefully crafted to drive advancements in machine learning across various applications. This dataset offers a rich exploration into the world of upright piano music, capturing the timeless elegance and expressive depth of this iconic instrument.

At the core of the Upright Piano Dataset is the distinctive sound of the upright piano, revered for its warm resonance and intimate character. From delicate melodies to thunderous chords, the upright piano offers a versatile canvas for musical expression across a wide range of genres and styles. Each audio track within this dataset showcases the unique timbre and nuances of the upright piano, providing a wealth of material for exploration and analysis.

Accompanying the audio tracks are detailed metadata annotations, offering insights into the musical structure, instrumentation, key signatures, tempo variations, timestamps, and more. This comprehensive metadata empowers machine learning models to comprehend the intricacies of upright piano music composition and performance, facilitating tasks such as generative music composition, music information retrieval (MIR), source separation, and beyond. The Upright Piano Dataset serves as a valuable resource for researchers, musicians, and developers interested in harnessing the expressive power of the upright piano within the realm of machine learning, unlocking new avenues for creativity and innovation in the dynamic landscape of music synthesis.