Jose Ranero GarcíaORCID: 0009-0009-3168-6379 License: CC BY-NC 4.0Date: June 2026 Companion record: Argira Sonification Pipeline v3.4 (DOI: 10.5281/zenodo.20550714) Description Argira Canonical Corpus v2.0 extends the visual corpus used by the Argira Sonification Pipeline v3.4 from 187 to 227 images through the addition of 40 canonical artworks from nine previously absent or underrepresented artists. The resulting corpus contains 96 canonical paintings, of which 75 are currently assigned to 19 identified artists, plus 131 synthetic control images used exclusively for pipeline validation. The dataset was developed to investigate whether structural visual metrics associated with sonification remain stable across artists, styles, and historical periods. All visual descriptors and sonification parameters were computed using the Argira Sonification Pipeline v3.4. The pipeline implements a three-channel orthogonal architecture: Channel A: hue_entropy → base frequency Channel B: edge_density + fractal_D → modulation depth (OLS-derived predictor, R² = 0.640) Channel C: luminance_contrast → decay time Main Contributions The expanded corpus reveals several artist-level structural signatures in the joint space defined by fractal_D and edge_density: Turner occupies a stable high-fractal / low-edge region across multiple works and periods. Klimt exhibits a bimodal distribution, spanning both low-edge and high-edge regimes. Caravaggio concentrates in a low-fractal / medium-edge zone consistent with large dark regions and localized contour structure. Rembrandt maintains relatively stable fractal structure while varying substantially in edge density according to composition. Degas shows consistently high fractal dimensions combined with a wide range of edge densities. These observations suggest that artist-level organization emerges naturally in fractal_D × edge_density space without explicit stylistic labels. Structural Convergence A central finding of the corpus is the presence of structural convergence: visually distinct paintings frequently occupy nearby positions in the predictor space used by the sonification system. The convergence phenomenon previously documented in Argira v3.4 generalizes to the expanded corpus. Eighty pairs of canonical paintings satisfy: |Δ fractal_D| < 0.005 |Δ edge_density| > 0.15 yet produce nearly identical modulation-depth values. This result indicates that substantially different visual organizations may converge toward similar acoustic states within the Argira mapping architecture. Across artists, mean fractal_D spans from 1.633 (Caravaggio) to 1.878 (Kandinsky), suggesting that fractal structure captures long-range stylistic differences not fully reflected by edge density alone. Artists traditionally associated with atmospheric or abstract visual organization (Turner, Monet, Kandinsky) occupy systematically higher fractal_D regions than artists centered on localized figure-ground contrast (Caravaggio, Rembrandt). This observation remains descriptive and motivates future investigation rather than constituting a formal historical claim. The Turner entry with edge_density = 0.000 corresponds to a synthetic control image and should not be interpreted as a canonical artwork. Corpus Composition Category N Previous corpus (v3.4) 187 New canonical works 40 Total images 227 Canonical paintings 96 Synthetic controls 131 New additions include works by Turner, Klimt, Degas, Velázquez, Caravaggio, Rembrandt, Cézanne, Goya and Delacroix. Dataset Statistics Canonical Paintings (N = 96) fractal_D: 1.588–1.916 (mean = 1.789) edge_density: 0.056–0.832 Full Corpus (N = 227) fractal_D: 1.116–2.000 (mean = 1.732) Twenty-eight canonical paintings (29%) exceed fractal_D = 1.85 and activate the granular synthesis layer of the pipeline. Files sonificacion_resultados.csv — Full dataset (227 images, 28 variables) corpus_v2_artist_summary.csv — Artist-level statistics fractal_vs_edge_canonical_v2.png — Scatter plot of canonical paintings in fractal_D × edge_density space sonify_painting_v3_4.py — Full sonification pipeline README_corpus_v2.md — Documentation Related Records Argira Sonification Pipeline v3.4 — DOI: 10.5281/zenodo.20550714 ARGIRA V: Cross-Corpus Evidence for Predictor Inversion in Visual-to-Acoustic Mapping — DOI: 10.5281/zenodo.20534328 ARGIRA Series (Concept DOI) — DOI: 10.5281/zenodo.20534327 Citation Ranero García, J. (2026). Argira Canonical Corpus v2.0 — 96 Canonical Paintings, 19 Artists. Zenodo.