FFT SPECTRUM
🎛️ Audio FX Modulation
✨ Visual FX & Hologram
🎥 Stage Camera & Export
The saved animation reads frequency bytes only while its analyser exists and playback is active. Low uses bins one, two and three, mid eight, nine and ten, and high sixteen, seventeen and eighteen. Each group is averaged and divided by two hundred fifty five. For illustrative low bytes zero, one hundred twenty seven and two hundred fifty five, the normalized value is three hundred eighty two divided by seven hundred sixty five, about point four nine nine three five. Compare all three bytes two hundred fifty five: normalized value one. Bars use four hundred pixels per normalized unit, about one hundred ninety nine point seven four and four hundred. These fixed index groups are source choices, not automatically identified instruments or genres. Actual bin frequencies depend on the analyser and audio context configuration. Hologram scale is one plus low frequency times point five times bloom intensity. For illustrative low point five and bloom two, scale is one point five. Ring scale instead is one plus mid frequency times point two; illustrative mid point five produces one point one. Bars use two hundred pixels per scale unit, widths three hundred and two hundred twenty. These are relative geometry scales, not measured brightness or audio loudness. Beam opacity is point one plus low times point four, halved for the additive term on odd beam indices. At low point five, even beams get point three and odd beams point two. Time-based rotation and camera movement continue separately from this audio response. Export JSON writes the entire configuration state with a title, version and timestamp. It does not encode the generated audio, an FFT recording or a video. The illustrative diagram counts one configuration object and zero recorded media objects, at three hundred pixels per object. The saved boot sequence initializes Three.js before binding interface controls. In isolated offline checks, a missing Three.js dependency can therefore prevent later bindings. That must be compared against the original before claiming preserved functionality. Source references to a named label or artist do not prove authentic recorded material. Local narration for this explainer is separate Kokoro speech, and actual tool sound, responsive canvas buffers and public playback need their own evidence.