Entropy Extraction from Audio Signals Using Hilbert-Mod Transformation and Von Neumann Refinement

Özet

This paper introduces a multi-stage framework for extracting high-entropy binary sequences from audio signals. The procedure consists of generating the an alytic signal via Hilbert transformation, applying thresholding and amplification, and performing modular reduction before binary encoding based on local ampli tude comparisons. Experiments were carried out using several prime moduli and amplification parameters, with optional post-processing through XOR aggregation and Von Neumann debiasing. Randomness quality was assessed using the NIST SP 800-22 test suite. The results indicate that configurations employing larger modulus values yield statistically robust bitstreams, while smaller moduli tend to introduce detectable structural patterns. Overall, the method provides effective entropy enhancement, particularly when dispersion and debiasing parameters are appropriately tuned.