This is the actual tool (rebuilt for the browser) that was used to manually verify the R‑peaks of all 1,920 ECG trial segments in our NeuroIS deepfake experiment — here loaded with eight real segments from the study's cleanest recording. Judge the rhythm, correct the automatic detection wherever it erred, and press save (s) or Next ▶ to check a trial off. A click on the signal adds or removes a peak; the strip at the bottom shows the entire segment, and dragging its box moves the 10‑second working window — drag or scroll on the traces themselves to pan and zoom. All keyboard shortcuts (a / d, f, r, u, s, n, arrows) work right away.

Real study data (pseudonymous participant IDs; ECG recorded at 1 kHz, drawn here at 250 Hz; the untouched original files are in segments/ and corrected_peaks/). The inspection is blind: participant, round, experimental condition and behavioural outcome are never displayed, so the judgement rests on the signal alone. Automatic detection: NeuroKit2 with artifact correction. Template matching, gap suggestions, windowing, tolerances and shortcuts behave exactly as in the Python original; panning, zooming and the overview strip are additions of the browser rebuild.