Glossary — FROID

IPM — Multimodal Power Index: The "speedometer" of session energy: quantifies, from 0 to 100, the patient's expressive intensity and psychomotor vitality, fusing voice, face, and semantics. A sustained drop suggests lethargy; spikes suggest hyperactivation.
IDM — Multimodal Deviance Index: The "compass" of coherence: assesses whether the patient's reactions are congruent with the content. Points out incongruities between speech and facial expression (dissonance), signaling emotional "hot spots."
MFCC — Mel-Frequency Cepstral Coefficients: A mathematical representation of vocal timbre, widely used in speech processing. FROID tracks MFCC7 and MFCC9 and their variation over time.
MFCC7 / MFCC9: Specific coefficients that, in the anchor study (Zhao et al., 2022), were associated with depression (MFCC7) and somatic anxiety (MFCC9).
Δ and ΔΔ (delta and delta-delta): First and second time derivatives of a coefficient — the "speed" and "acceleration" of change. ΔΔMFCC9 is read as the acceleration of laryngeal tension.
F0 — Fundamental Frequency: The pitch of the voice. Lower F0 variability (monotone speech) is a studied marker of psychomotor retardation.
ZCR — Zero-Crossing Rate: How many times the audio signal crosses zero per unit of time; a simple indicator of frequency content and noise.
Jitter / Shimmer (proxy): Frequency instability (jitter) and amplitude instability (shimmer) of the voice. In FROID these are normalized proxy indices, not the classic clinical Praat values.
HNR — Harmonic-to-Noise Ratio: Measures how much of the voice is harmonic sound versus noise. Used as a Quality Gate: low HNR indicates degraded audio and reduces the vocal channel's reliability.
FACS — Facial Action Coding System: The gold-standard system for describing facial movements in objective units (Action Units), independent of emotional interpretation.
AU — Action Unit: Each facial muscle or muscle group coded by FACS. E.g., AU6 (contraction around the eyes) and AU12 (raising of the mouth corners).
Onset · Apex · Offset: The three temporal phases of an expression: start, peak, and end. Tracking them helps distinguish spontaneous expressions from voluntary ones.
FFT — Fast Fourier Transform: An algorithm that decomposes sound into its component frequencies. It is the physical basis behind the spectral energy used in the 12 Zones.
Dynamic Baseline (60s calibration): The patient's "biological fingerprint," measured in the first minute of each session. Every subsequent reading is a deviation relative to it — the main bias-mitigation mechanism.
Clinical Screen Stabilization: A proprietary module that separates processing from presentation: FROID keeps processing at a 1-second cadence, but the screen shows a configurable clinical window (real time, 1, 3, 5, or 7 min; default 5 min) so the professional has time to assess all metrics. Critical alerts break the stabilization and update immediately.
Moving clinical window / Technical micro-window: Every 1 minute, FROID closes a technical micro-window; the screen consolidates the last 5 by temporal weighted average (10% · 15% · 20% · 25% · 30%, from the oldest to the current minute) — visual stability without losing the gradual reaction to the present.
Peak lock: A consolidation rule for clinical risks within the stabilized window: a recent critical peak does not immediately disappear from the screen, even if the weighted average has already dropped.
D_energy — Relative Energy Deviation: How far a zone's current energy deviates from the patient's own baseline. Ensures that stable biological traits are not read as symptoms.
M_fac — Facial Dissonance Multiplier: A factor (up to 2.5) that amplifies the IDM when the face contradicts the voice — for example, a voluntary smile (AU12 without AU6) during vocally tense content.
12 Perception Zones: A visual interface that distributes vocal energy across 12 bands associated with clinical themes. The physics is real; the theme↔band mapping is a product heuristic, not published neuroscience.
Somatoaffective Dissonance (False Calm): A state in which stress has been repressed from the face but keeps tensioning the larynx — an apparent calm that the IDM helps reveal.
Edge AI (Edge Processing): Audio and video are processed locally, on the practice's computer, without sending raw biometric files to the cloud.
k-anonymity (k ≥ 50): A technique that guarantees each anonymous record is indistinguishable from at least 49 others, making it infeasible to re-identify a patient in the Data-Froid.
Anonymous Data-Froid: A repository of de-identified session data used to audit bias and statistically validate FROID's heuristics at scale.

No term found. Try another word.

FROID's terms, in plain language