Video and audio — examples

Each example is a real job, built and run on Madoo on real media. The pattern is the one of the whole section: the AI listens, translates and speaks; the deterministic nodes measure, place, mix and check — how long a line lasts, where it goes on the timeline, how loud the result is, whether it can be delivered. Every example shows its results and its real cost, explains its nodes, and installs into your workspace with an API key.

Example	What it shows	Credits per run
W01 — A briefing dubbed into another language	Six minutes of NASA interviews, four people, dubbed into Italian: one transcription with the speakers, a voice cast per speaker, the translation adapted and voiced in up to three rounds until each line fits, the background kept, subtitles in both languages, a manifest and a quality report; the language is an input	≈7 per minute of video
W02 — A screen recording, finished for publishing	A one-minute product walkthrough from a laptop microphone: the voice cleaned and levelled, music written by an AI that watches the video, fitted and ducked under the voice, −14 LUFS, subtitles that highlight each spoken word, an SRT and a delivery report; every stage its own output	≈3 for a one-minute recording
W03 — A long recording, finished for publishing	The finish of W02 set for half an hour to two hours: the brief from an excerpt, a three-minute looped bed, transcription in segments billed under ten minutes each and rejoined, block subtitles; rehearsed on two and a half minutes with the long paths forced	≈4 plus 0.5 per minute

To install an example, see Installing an example: the same installer works for these workflows (--from the documentation site, or a local folder).