Record, or upload what you already have
Sixty seconds is enough to start; up to twelve minutes gives a fuller reading. An ordinary laptop or phone is fine, no studio, no wearables, no second camera. You consent explicitly before anything is analysed.
Four channels are measured independently
Words, voice, face and body are each analysed on their own terms, so a weak signal in one never quietly contaminates another. Anything that cannot be measured, a call with no video, a face turned away, is recorded as absent, never scored as zero.
Hundreds of parameters resolve into five dimensions
Clarity, conviction, energy, warmth and confidence. Each is built from measurements you can inspect individually, and each is reported with how much of the recording actually supported it.
The channels are checked against each other
The congruence score asks whether what you said, how you said it, and how you looked while saying it agreed with each other. Agreement across channels is what an audience reads as conviction; disagreement is what they read as something being off.
You get a distance, not a verdict
The report opens with the one thing to work on and the gap to your benchmark. The score comes third, deliberately, a score on its own changes nothing.
What SPECTRIA does not measure
These are not gaps waiting to be filled in a later version. They are claims the instrument cannot support, so it does not make them.
Honesty or deception
No claim is made about truthfulness. Delivery is not a polygraph.
Personality or traits
Nothing here describes who someone is, only how one recording landed.
Intelligence or ability
A delivery measurement is not a competence measurement.
Emotion as ground truth
Facial movement is reported as movement, a smile that reached the eyes is described as exactly that, not as happiness.