video-cleaning

Removing pauses and filler words from Korean video uses Whisper for word-level transcription to identify dead air and filler words like 어, 음, 아, then FFmpeg cuts those segments while preserving meaning. This Skill automates that workflow with conservative editing rules designed for lectures, podcasts, and interviews.

Word-level timestamps enable precise editing by pinpointing exactly when each word occurs. Whisper transcription generates these timestamps, which this Skill uses to identify and remove long pauses and Korean filler words with timing accuracy, producing both a cleaned video and detailed edit reports.

The workflow is: transcribe your video with Whisper to get word-level timing, preview the proposed edits to verify segments before cutting, then generate the final edited video using FFmpeg. This Skill handles all three steps, removing dead air while keeping content intact.

This Skill is designed specifically for Korean-language presentations, lectures, podcasts, and interviews. It recognizes Korean filler words and applies conservative editing rules tailored to Korean speech patterns, making it suitable for content in that language.

The editing process produces three outputs: a cleaned video with pauses and fillers removed, word-level transcriptions with timestamps, and a detailed edit report showing time savings and segment recaps. This gives you both the final video and documentation of what was changed.

Conservative editing preserves natural speech flow and meaning by removing only clear dead air and recognized filler words, avoiding over-editing that damages content. Use this approach for professional videos like lectures and interviews where maintaining integrity matters more than maximum length reduction.