Autonomous Media & Tutorial Studio

Vibe-Cut Studio

An agent-driven video editor. Describe the cut in plain English, an AI writes the JSON shot list, FFmpeg renders it in both 16:9 and 9:16 — no timeline, no mouse, no editor.

Inspect Media Pipeline ← Back to Hub
Precision Slicing

FFmpeg Lossless Trim

Every shot is an in/out timecode in a JSON file. Change one number, re-render. Audio levels and scene-change detection tell you where the dead air is.

AI-Readable Footage

Contact Sheets

sheet turns a recording into a grid of timecoded frames an AI can actually read — so the model picks the cuts, then sets pan and zoom per shot in the shot list.

Studio Narration

25+ Natural Voices

Whisper transcribes your recording to timecoded SRT. Narration is synthesised locally from 25+ natural voices — nothing uploaded, nothing billed per word.

Delivery

Captions & Aspects

Burns titles and captions straight into the render, and outputs 16:9 and 9:16 from the same shot list — one source, every platform.

The Zero-Friction Production Flow

Record once. Hand the footage to an agent. Get a finished cut back.

vibe-cut — automated media pipeline
[1. Ingest]          Probing raw screen recording: recording_raw.mp4 (1080p, 60fps)
[2. Sheet]           Contact sheet written: 41 timecoded frames for the model to read
[3. Shotlist]        Agent wrote 9 shots • in/out timecodes, captions, zoom per shot
[4. Voice]           Local TTS narration rendered • voice: bm_george
[5. Render]          FFmpeg filtergraph • zoompan, burned captions, 16:9 + 9:16

[✓] Rendered: tutorial_final_16x9.mp4 + tutorial_final_9x16.mp4