Argira Phase 3 — Preliminary Perceptual Test Results (Pilot, N=5)

Argira v3.4 introduces an extended statistical analysis of the relationship between visual fractal dimension (Dv), chromatic variability (hue_std), and the spectral fractal dimension of the sonified spectrogram (Ds). The goal is to determine whether color affects Ds and whether Dv contributes independently to the structure of the spectrogram. Pipeline stabilityThis version maintains the optimized STFT configuration introduced in v3.3 (Hann window, 1024‑segment, 768‑overlap, 2048‑FFT), logarithmic spectrogram scaling (log1p), adaptive binarization (80th percentile), and fixed box-counting scales (4, 8, 16, 32, 64). These parameters ensure stable and reproducible Ds values across artworks. DatasetThe dataset includes 20 paintings from different authors and periods, processed through the Argira sonification pipeline. Simple correlations- Dv vs Ds: r = –0.0146 - hue_std vs Ds: r = 0.8371 - tempo vs Ds: r = 0.2621 Color is the dominant predictor of Ds. Partial correlationControlling for hue_std: - Partial correlation Ds–Dv | hue_std: r = 0.6989 This reveals a suppression effect: the influence of Dv on Ds is hidden by the strong effect of color. Multiple regressionModel: \[Ds = \beta0 + \beta1 \cdot hue\std + \beta2 \cdot Dv + \beta_3 \cdot tempo\] Results: - β(hue_std) = 0.3493 (t = 8.993) - β(Dv) = 0.2206 (t = 3.690) - β(tempo) = –0.0001 (t = –0.319) - R² = 0.8479 - R² with hue_std only = 0.7008 - Increase when adding Dv = +0.1471 Interpretation: Chromatic variability explains ~70% of the variance in Ds. Dv contributes an additional 14.7%, and its coefficient is statistically significant. Tempo does not meaningfully influence Ds. ConclusionpgColor strongly affects the spectral fractal dimension (Ds), but visual fractality (Dv) also contributes significantly once the chromatic effect is controlled. The relationship between Dv and Ds is real but partially masked by the dominant influence of color. These findings support the hypothesis that Argira captures emergent crossmodal correspondences between visual and auditory complexity. Included files- analyce_folder_v3.py - results.jpg- README.md LicenseCC BY‑NC 4.0 AuthorJose Ranero García (Argira) ORCID: 0009‑0009‑3168‑6379