The speech simulator uses a keyword heuristic, not clinical sentiment assessment. Each different positive keyword found as a substring contributes point four; each different negative keyword subtracts point four. Repeated appearances of the same word do not add again because each keyword is tested once. Two positive keywords give point eight; three give one point two before the result is clamped at one. The bars represent the two unclamped magnitudes at three hundred pixels per score unit, reaching two hundred forty and three hundred sixty. If positive and negative terms exactly cancel, the code uses a length fallback: text longer than thirty characters gets point two. This rule can label a balanced long phrase positive without understanding its meaning. The labels called vagal tone, heart rate variability and cortisol are calculated numbers, not measured biomarkers. For an illustrative valence of negative point six, the source rounds point five five plus point three five times valence to point three four. Its heart rate variability label is twenty plus sixty five times point three four, rounded to forty two. For valence point eight, the first label is point eight three and the second rounds to seventy four. The two bars show the synthetic first label at four hundred pixels per unit, giving one hundred thirty six and three hundred thirty two pixels. These equations do not establish physiological causation, diagnosis or treatment benefit. The source also computes spectral energy as point three plus point five times one minus its synthetic tone label. At point three four that yields point six three; at point eight three it rounds point three eight five to point three nine. The two bars share four hundred pixels per unit, reaching two hundred fifty two and one hundred fifty six. These are modeled values rather than recordings of a person. The external p five library is unavailable in this offline preview. Sketch construction occurs before listener binding, so the original page cannot prove working controls, drawing or exports under that condition. The explanation preserves that limitation. No medical effect, actual voice analysis, successful export or biological improvement is claimed.