Planning
Reelery
Write it. Watch yourself say it.
Local-first video authoring studio. A written script becomes a finished video — narrated in your own cloned voice, presented by your own face, captioned and cut — without the source material leaving your machine.
A proposed product, not a released tool. Features below describe the intended direction.
View the changelog →Setup & access
No usable release is established by this catalog entry. Do not treat the feature description as installation instructions.
Vision
Hosted avatar platforms charge between $2 and $10 for every finished minute of video. The same minute rendered on a desktop GPU costs about three cents of electricity. Reelery exists to close that gap without giving up quality — and without handing the raw material to a third party.
The bet
The realism ceiling on a synthetic presenter is set by the source material, not by the model. A few minutes of recorded footage turns every future video from a generation problem into a lipsync problem — cheaper, faster, and markedly more convincing, because most of every frame is real video.
Reelery is built around that: record yourself once, then write.
Scope
Script → per-beat voice-over → locked audio timeline → presenter render → word-level captions → assembly with b-roll and titles → a QC gate that actually inspects the output → publish with the AI marking and disclosure the law now requires.
Dictation is deliberately out of scope; that belongs to its sibling, Dictaro. The two share the voice identity, the speech-recognition worker and the media conventions.