Automatic Music Transcription (AMT) remains a challenging task in music information retrieval (MIR), particularly for polyphonic piano music where multiple notes overlap in time and frequency. This paper presents Piano2Notes , an end-to-end modular system that converts raw piano audio into structured digital notation (MIDI and staff notation). The system combines classical digital signal processing (DSP) for onset detection with a convolutional recurrent neural network (CRNN) for multi-pitch estimation, followed by a Hidden Markov Model (HMM) for note tracking. We evaluate Piano2Notes on the MAESTRO dataset, achieving a frame-level F1 score of 0.89 and a note-level F1 of 0.83. The system is optimized for real-time performance on consumer hardware and provides an accessible interface for musicians and educators.
The world of music has undergone a significant transformation in recent years, with the rise of digital tools and software that enable musicians to create, record, and produce music with unprecedented ease and flexibility. One such innovation that has gained immense popularity among music enthusiasts is the Piano2Notes mod, a revolutionary software that allows users to convert piano recordings into sheet music. piano2notes mod
Raw Audio → Preprocessing → Onset & Pitch Detection → Note Tracking → MIDI/Sheet Generation We evaluate Piano2Notes on the MAESTRO dataset, achieving