07 — Scalograms: seeing frequency over time

flowchart LR
    SIG["beat window"] --> CWT["CWT: slide Morlet probe\nstretch = low freq\nsqueeze = high freq"]
    CWT --> HEAT["heatmap\ntime left-right\nfrequency bottom-top\nbrightness = energy"]

Fourier says which frequencies exist but not when. The scalogram keeps both: QRS is a bright vertical streak (high frequencies, one instant); P and T glow lower and wider.

• QRS spike: bright vertical streak, high up — sharp means high frequency, brief.
• P wave: dim wide glow, low — gentle means low frequency.
• T wave: medium glow, low-middle — a recovery slope.
• V beats: smeared bright bands — wide bizarre shapes spread energy.

Shape ≈ 32 frequency rows x 216 time columns — a small image, read by a 2D network (lesson 08). Research question 1 tests whether this view beats raw signal alone.

Next: 08-neural-nets-cnn-lstm.md