How automated direction shows up in your film.

A short cinematic film is a composition. It needs an emotional arc, a soundtrack that breathes, narration that earns its place, and visual pacing that gives the room its own room. Souvenyr's direction layer makes those calls quietly, using the atmosphere you described at intake and the contributions your guests left.

• Reads the contributions to identify the emotional beats of the day — the speeches, the reactions, the still that catches the breath between two louder ones.
• Composes a soundtrack appropriate to the mood you set, chapter by chapter, then renders it as one piece of music. Narration is read by the voice you choose, in the language you choose.
• Places voice memories at the right moment in the film, attributed to the guests who recorded them — never moved into anyone else's mouth.
• Adds cinematic light and ambient transitions in a few places, sparingly, where the film genuinely benefits from them. Never as a substitute for the day.

• It never generates faces, identities, or people that weren't there.
• It never invents events, dates, names, or quotes you didn't provide.
• It never publishes for you. Every share is your decision.
• It never trains on your event. Your media is not enrolled in any training set, by us or by any partner we work with.
• It never overrides the editorial standards. Restraint is the point.

A short cinematic film about the most important day of someone's life can become saccharine very quickly. Souvenyr resists that by default. Lines like "fairy-tale," "once-in-a-lifetime," "tears of joy," and "magical evening" are removed before they ever reach the narrator. The film either earns the line or holds in silence.

The same standards apply on every plan, every event, every theme. Narration appears on at most 18% of beats — the room is allowed to speak for itself. Consecutive emotional peaks are capped at two — the film breathes between its loudest moments. Stylised cinematic inserts are capped at three — imagery is used like seasoning, not the dish. These are not tunable. They are the editorial floor.

You can disable narration entirely. You can choose a different soundtrack direction. You can request changes on the storyboard before render. You can move a film back to draft after it ships. Every step is reversible until you choose to share.

Voice memories are uploaded by named guests, played back attributed, and never put into anyone else's mouth. If you ever ask us to remove a voice memory, we delete the recording and re-compose the film without it. The original speaker can also request removal directly.

Souvenyr's direction layer is composed of multiple specialised systems contracted under data-processing agreements. We deliberately don't dwell on which model produced which sentence; the point of the film is the day, not the supply chain. If you need a list of subprocessors for your own compliance review, write to policy@souvenyr.film and we'll provide it.

Write to policy@souvenyr.film with any concern about how automated direction is used in your event. We respond within one business day.