video_edit_skill

Overview

This skill performs intelligent editing of recorded vlog footage to remove repeated lines, clear mistakes, and preserve natural flow. It uses speech transcription and an alignment with your script to generate a precise cut plan, then applies edits after user confirmation. The result is a natural, concise vlog with preserved emotion and smooth transitions.

How this skill works

The skill extracts audio from the input video and transcribes speech with time-stamped segments. It compares the transcript to your provided script to align spoken content, detect repeats, misreadings, long pauses, and filler words. It produces a cut_plan.json listing keep and removed segments, presents the plan for user review, and executes frame-accurate cuts with configurable buffers.

When to use it

• Editing solo-hosted vlog recordings to remove mistakes and repeats
• Cleaning talking-head footage where you want natural pacing preserved
• Preparing social media clips from longer raw recordings
• Batch-processing multiple takes to keep the best performance
• When you have a reference script to align against for accuracy

Best practices

• Provide a clear reference script to improve alignment accuracy
• Review suggested removals before finalizing to keep intended personality
• Keep short thinking pauses (1–2s) to preserve natural flow; remove long pauses (>3s)
• Allow small buffers (0.1–0.3s) around kept segments to avoid jarring cuts
• Ensure the transcription service and ffmpeg are reachable and up to date

Example use cases

• Trim a 20-minute raw vlog to a focused 8–10 minute episode by removing repeats and long pauses
• Generate short highlight reels by keeping only key topic segments identified by script alignment
• Remove flubbed lines and rejoined takes while preserving laughter and emphasis
• Automate editing for a recurring series where format and script are consistent
• Create polished clips for social platforms from longer recorded talks