Raw experimental data for "Exploring High-Dimensional Spaces: Evaluation of a Node-Based Interface for Neural Audio Synthesis"

Raw experimental data accompanying the paper Exploring High-Dimensional Spaces: Evaluation of a Node-Based Interface for Neural Audio Synthesis. Sixteen participants (p01–p16) each built a musical-instrument mapping over a 128-dimensional latent space controlling a real-time neural audio decoder, twice: once with the proposed node-based interface and once with a baseline of one slider per dimension, in counterbalanced order. Each session was followed by questionnaires. Contents experiments.zip — per-participant session records: committed presets (128-dimensional vectors and node layouts), in-session preset screenshots, demonstration data logging every latent-space position the participant moved through, per-condition timings, session metadata, and questionnaire responses (Goldsmiths Musical Sophistication Index; Creativity Support Index, first, second and paired). preset-screenshots.zip — rendered scene screenshots of each participant’s committed presets, named by interface. README.md — full description of the directory layout, file formats, and field semantics. Read this first. Not included: the 1.6 GB summed sine (mixtones) training dataset, which is available from the supplementary site at https://appendix.leoauri.com/ehds/#dataset. Participant demographics were collected separately and are reported only in aggregate in the paper. Anonymisation: participants are identified only by the pseudonymous ids p01–p16. No names, contact details, or free-text responses are present; all questionnaire responses are numeric.