INPUT
The test samples a seven-second live microphone stream while you read one fixed sentence. The stream is processed by the Web Audio API in your browser; no recording file is created. Very quiet frames and pitch estimates outside the expected 70–280 Hz analysis range are rejected.
THE FOUR GAMEPLAY FEATURES
- Depth, 38%: primarily derived from median fundamental frequency.
- Resonance, 26%: a rough high-frequency-content proxy, not a clinical formant analysis.
- Weight, 18%: a normalized signal-energy proxy affected by microphone distance and gain.
- Control, 18%: pitch stability across accepted frames.
The weighted score is clamped to 31–97. Age selection labels the comparison view but does not change the acoustic score in V.04.
PERCENTILE AND ARCHETYPE
The percentile is a logistic mapping of the Voice T Score. It is not calculated from a representative U.S. sample. Archetypes are entertainment labels based on fixed score bands. They should not be interpreted as biological categories.
WHY THESE CUES?
Speech-perception research consistently finds that median fundamental frequency is an important cue in perceived vocal masculinity, while resonance and other acoustic features also contribute. A 2024 study found that pitch was the primary cue but that accurate perceptual models require additional acoustic properties. A 2025 study also emphasized substantial variation across listeners and groups.
THE HARD LIMIT
Correlation does not make an individual voice recording a blood test. Adult voice is also shaped by anatomy, puberty, age, learned behavior, language, illness, recording conditions and deliberate vocal technique. High T Voice therefore makes no claim about current testosterone, medication use, fertility or health.
BETTER PERSONAL COMPARISONS
For a more useful runback, use the same device, room, sentence, distance and time of day. Personal trends under controlled conditions are more defensible than comparing absolute scores across different hardware.