Keyboard Music Patterns by Combining Mathematical Analysis and AI Tools

Kashish Nair

Abstract: This study explores keyboard music patterns by combining mathematical analysis and AI tools. Over six weeks, we collected scale and chord data, mapped notes to numerical representations, and used Python libraries (music21, mido, NumPy) to quantify intervals, chords, and rhythm. We then generated new melodies using Google?s Magenta model. Our findings reveal consistent intervallic and rhythmic structures in human-played scales and chords, in line with music theory. AI-generated melodies exhibit similar large-scale tonal movement but differ in micro-timing and harmonic detail. This study demonstrates that AI tools can replicate basic musical structures, though they have limitations in nuanced expression. The study investigates whether AI-generated melodies preserve the statistical and intervallic properties of human-played keyboard scales. It is hypothesised that AI models will reproduce macro-level tonal structure but diverge in micro-timing and harmonic detail.

Keywords: MIDI data, interval patterns, rhythm quantification, Markov chain modeling, AI-assisted music generation, Google Magenta, MelodyRNN, music theory analysis, computational musicology, mathematical pattern detection

How to Cite?: Kashish Nair, "Keyboard Music Patterns by Combining Mathematical Analysis and AI Tools", Volume 14 Issue 11, November 2025, International Journal of Science and Research (IJSR), Pages: 1973-1977, https://www.ijsr.net/getabstract.php?paperid=SR251129154230, DOI: https://dx.doi.org/10.21275/SR251129154230