How dancan254/voiceover-video-skill works
voiceover-video is an agent-orchestrated, local pipeline for turning a recording—or an agent-written script—into an edited vertical video. The principal workflow transcribes and proofreads speech, builds word-timed captions, plans and approves shots, sources credited media, authors themed scenes, renders and reviews frames, then synthesizes sound and encodes the final video. Script-mode voices and aligns character dialogue instead of transcribing an input recording. The README documents additional features such as face-cam bookends and alternate formats; sampled source confirms only selected implementation details, so uncertain internal wiring is omitted.
Main entry points: render.js Frame renderer · mix-encode.sh Mix and encode · find_media.py Media search and fetch · kinetic.html Scene composition
Components 21
Workflow and speech
- Workflow instructions
skills/voiceover-video/SKILL.md - Voice recording
- Local transcription
skills/voiceover-video/scripts/transcribe.py - Script voiceover
skills/voiceover-video/scripts/speak.py - Caption preparation
skills/voiceover-video/scripts/build_captions.py
Media and scenes
- Approved shot plan
skills/voiceover-video/examples/kafka-vs-rabbitmq/shot-list.md - Media search and fetch
skills/voiceover-video/scripts/find_media.py - Media providers
- Media credits
skills/voiceover-video/scripts/find_media.py - Scene composition
skills/voiceover-video/templates/kinetic.html - Brand configuration
skills/voiceover-video/scripts/init_brand.py - Theme template fill
skills/voiceover-video/scripts/fill_template.py
Rendering and review
- Frame renderer
skills/voiceover-video/scripts/render.js - Contact sheet review
skills/voiceover-video/scripts/contact-sheet.sh - Frame encoding
skills/voiceover-video/scripts/render-frames.sh
Sound and delivery
- Timeline sound cues
skills/voiceover-video/scripts/render.js - Sound synthesis
skills/voiceover-video/scripts/synth_audio.py - Mix and encode
skills/voiceover-video/scripts/mix-encode.sh - Finished video
Other components
- Creator
- Shell-capable agent
Connections 29
- Creator → Shell-capable agent: requests video
- Shell-capable agent → Workflow instructions: follows workflow
- Creator → Voice recording: provides recording
- Voice recording → Local transcription: transcribes locally
- Local transcription → Caption preparation: writes word timings
- Shell-capable agent → Caption preparation: proofreads and builds
- Shell-capable agent → Script voiceover: voices script
- Script voiceover → Caption preparation: writes aligned words
- Shell-capable agent → Approved shot plan: authors and approves
- Shell-capable agent → Media search and fetch: searches and fetches
- Media search and fetch → Media providers: queries providers
- Media search and fetch → Media credits: records attribution
- Shell-capable agent → Scene composition: authors scenes
- Creator → Brand configuration: sets brand
- Brand configuration → Theme template fill: provides configuration
- Theme template fill → Scene composition: fills composition
- Frame renderer → Scene composition: loads composition
- Scene composition → Frame renderer: provides timeline
- Frame renderer → Contact sheet review: renders review stills
- Shell-capable agent → Contact sheet review: reviews and revises
- Frame renderer → Timeline sound cues: extracts cues
- Timeline sound cues → Sound synthesis: supplies timeline cues
- Frame renderer → Frame encoding: renders frames
- Frame encoding → Mix and encode: provides rendered video
- Sound synthesis → Mix and encode: provides sound beds
- Voice recording → Mix and encode: supplies voice audio
- Media search and fetch → Mix and encode: supplies extracted clip audio
- Mix and encode → Finished video: encodes final video
- Finished video → Creator: delivers result