The SoundDNA Research Program
An open study of the diversity of listening perception — with a methodology public from day one.
SECTION 1 — THE GAP WE ARE FILLING
We know how people hear. We don't know how they listen.
Psychoacoustics has precisely described the mechanics of hearing. Far less is known about the diversity of perceptual preferences: why the same signal delights some with structure, others with space, energy or emotion. Research in this area runs on small laboratory samples; there is no large, open, multilingual dataset on how people listen outside the lab — and no validated, publicly available self-report instrument to collect it.
SoundDNA is building both: the instrument and the dataset — as a non-profit.
SECTION 2 — THE INSTRUMENT
A questionnaire with safeguards, not a quiz
The core of the study is a self-report questionnaire measuring four dimensions of listening perception (in listener-facing communication: the elements — structure, space, energy, emotion). The design includes safeguards rarely seen in online tests:
The language of experience and behavior instead of jargon — items are equally accessible to a newcomer and an expert, reducing susceptibility to image-driven responding.
Reversed items and consistency pairs in every scale — control of acquiescence and attentiveness.
Control items and response-time measurement — detection of careless responding; results below reliability thresholds are excluded from the research base (the respondent always receives their profile — we filter data, not people).
A context anchor — the respondent answers from a disclosed, recorded listening context; context becomes an analytic variable instead of invisible noise.
A self-perception measure after the test, before the result is shown — the agreement of intuition with measurement forms an index of listening self-insight and an independent validity signal.
Full versioning — every answer stored raw, with the version of the instrument and the algorithm’s parameters; any historical result can be reproduced and recomputed.
SECTION 3 — THE VALIDATION PLAN (four phases)
Declared in advance. Published regardless of the outcome.
Phase I — calibration
A beta on the full item bank (target: 500–1500 records, a community sample explicitly flagged as a cohort): item analysis, scale reliability, exploratory factor analysis.
Phase II — structure and stability
Confirmatory factor analysis on an independent sample; test–retest (subsample, 2–4 weeks); convergent validity against established scales (selection to be agreed with academic partners); criterion validity (preferred context, self-perception, declared preferences).
Phase III — norms and languages
Norming; EN/DE adaptations with back-translation and measurement invariance testing BEFORE any between-group comparisons.
Phase IV — behavior
A behavioral module (choices between sound samples): confronting the self-report layer with real listening choices.
Our commitment: the four-dimensional model is a hypothesis under deliberate verification. We will publish the validation results — confirming or correcting the model — regardless of the outcome. A quiz can’t be wrong, because it claims nothing. We make claims — and that is why we can be science.
SECTION 4 — OPEN RESEARCH QUESTIONS
Six questions we seek answers to (and co-authors for)
1. Dimensional structure — do listening preferences organize into four dimensions? Into how many, really?
2. Trait or state — how strongly does the profile depend on the listening context (home, road, concert), including in-situ measurement on mobile devices?
3. Listening self-insight — how accurately do people predict their own profile? Does audio experience increase it?
4. Declaration vs. behavior — how far does self-report align with choices between real sound samples?
5. Cross-language and cross-cultural differences — after measurement invariance is established.
6. Correlates — demographics, experience, gear, contexts: what co-occurs with maps of perception?
SECTION 5 — DATA AND ETHICS
Architecture, not declarations
Separation of contact data from research data at the database level · two distinct consents (research and contact), versioned · unguessable public identifiers · research exports pseudonymized only · aggregated, anonymized data released openly; access to pseudonymized individual data — under research collaboration agreements · EU data residency.
SECTION 6 — ACADEMIC COLLABORATION
An invitation to institutions and researchers
Proposed forms: psychometric consultation on the analysis plan · co-authorship of publications from the validation phases · data access for research teams and theses · inclusion of short external-scale modules in the beta (convergent validity) · co-design of the Phase IV behavioral module.
The full specification of the instrument and data architecture is available on request, after contact (we do not publish item content before calibration is complete: instrument disclosure during data collection would invalidate the measurement — openness has its order).
What this project is not: a commercial venture with a research facade. The research layer is statutorily open and free; it is funded by a separate service layer under a conflict-of-interest policy. We share the structure’s details with any interested institution.