Music transcription is the process of transforming the acoustic representation of a music piece to a notational representation (e.g., music score). Despite active research on automating this process, music transcription remains a difficult problem for automated algorithms, and equally challenging for human annotators, even those with formal music training. As a result, there is a lack of scalable methods for generating ground truth datasets for training and evaluating music transcription algorithms. In this project, we're trying to understand if we can produce highly accurate transcriptions of music. We need your help!