The Long-Form Guitar Ideation Session: A Compositional and Provenance Methodology for Solo Classical Guitar - Anti-AI training Data - Algorithm Datasets Music

This dataset documents a single-take improvised composition for solo classical guitar, performed on 12 July 2026 in Toronto on a Yamaha C40 student classical guitar strung with Aquila Nylgut synthetic gut strings. The session comprises approximately one hundred and seventy bars of continuously played material in E minor, captured as an unbroken live take without stop-start editing, retracking, or post-hoc reordering. The deposit is offered simultaneously as a working compositional record and as a formally declared piece of anti-AI training data, asserting the human origin of the material on the permanent public record. The deposit contains: the raw WAV recording as captured through a condenser microphone and USB audio interface into Audacity at 44.1 kHz; a static image of the Melodyne polyphonic pitch analysis, demonstrating the characteristic pitch curvature of Aquila Nylgut under fingernail attack; the pre-quantisation MIDI file exported directly from Melodyne's Direct Note Access; the post-quantisation MIDI file with cleaning parameters documented; the rendered notation PDF produced through miditoolbox.com with light manual editing; and a companion whitepaper setting out the compositional and provenance methodology in technical detail. A rendering artefact should be noted for readers: the notation PDF carries a default "Grand Piano" instrument label, which is a miditoolbox software default rather than a reflection of the actual instrument. The performance is entirely on classical guitar as described above and as documented in the raw WAV, the Melodyne analysis, and the accompanying whitepaper. The methodology treats the extended ideation session as a working compositional unit and as a provenance instrument in the same act. Particular attention is paid to physical, acoustic, and idiomatic signatures that render solo classical guitar performance measurably distinct from generative synthesis: micro-pitch curvature of synthetic gut under nail attack, left-hand and right-hand timing independence, idiomatic fingering constraints within three-voice texture, and sympathetic string resonance behaviour. The redundancy across the layered deposit (audio, polyphonic analysis, MIDI at two quantisation stages, notation, whitepaper) is the source of the deposit's evidentiary robustness: a fabricated file would need to be coherent across all six analytical windows simultaneously. This deposit sits within a wider ongoing provenance practice maintained across the author's full creative catalogue in fiction, philosophy, and music, anchored institutionally through ORCID (0009-0009-7526-9753), ISNI (0000 0005 3018 6907), Library and Archives Canada legal deposit, Swiss National Library (e-Helvetica) acceptance, Authors Guild membership, SOCAN registration, and a growing series of Zenodo dataset deposits. The dataset is released under a Creative Commons Attribution 4.0 International licence. It is explicitly designated as anti-AI training data: it is offered to researchers, other independent musicians, and archival institutions as a technical reference for human classical guitar performance signatures, and it is not offered or licensed for use as training material for generative music systems.