A native macOS line reader and take recorder for replacing generated narration with a human performance. It opens a JSON screenplay, shows one line and its reference frame at a time, records 48 kHz mono WAV takes, and writes a resumable manifest.json containing every take and the selected take for each line.

narrator-wizard/bin/narratorwizard path/to/narration-spec.json

Keys: R records, S stops, P plays, and Return keeps the current take and advances. Closing the window is safe; the next launch resumes at the first line without a kept take.

The input menu lists live CoreAudio capture devices. After hot-plugging an interface such as a Focusrite, press Inputs, select it, and record. Optional input monitoring is off by default and should be used with headphones. The wizard restores the input that was selected before launch when it closes.

Appearance follows macOS Auto/Light/Dark and updates while the app is open.

When a screenplay supplies a video path, Play current cut opens that video inside Narrator Wizard using native AVKit playback, scrubbing, volume, and full-screen controls. This lets the narrator review the existing edit without leaving the recording session.

The Social Software reel is already wired:

narrator-wizard/bin/narratorwizard marketing/sosoft-reel/narrator-spec.json
node marketing/sosoft-reel/use-human-narration.mjs
node marketing/sosoft-reel/render-realtime-spine.mjs
node marketing/sosoft-reel/caption-and-mix.mjs
node marketing/sosoft-reel/export-delivery.mjs