A real-time AI supervisor that performs 8-band spectral analysis on every sample, classifies content, and applies bounded offsets to every processing stage — adapting the entire chain before problems become audible.
ProcIQ derives spectral centroid, tilt, flux, and energy ratios from an 8-band spectral analysis. Using a Neural Lite inference engine, it classifies every moment of audio as Voice, Music, or Mixed content. Classification drives bounded parameter offsets across every supervised processing stage — from AGC attack times to multiband thresholds to de-esser sensitivity.
ProcIQ never makes drastic changes. Every offset is bounded within safe limits, smoothed with configurable time constants, and tracked by a closed-loop effectiveness monitor. A minimum effectiveness floor of 30% prevents the system from fighting itself. The result is processing that sounds intentional, not reactive — listeners hear consistency, not correction.
ProcIQ provides full real-time telemetry — every classification decision, every offset applied, every effectiveness score. Engineers can monitor exactly what ProcIQ is doing and why. The reasoning output shows the spectral features driving each decision, making ProcIQ the most transparent AI supervisor in broadcast audio processing.
Hear how intelligent supervision keeps your processing consistent across every format and source.