What problem does it solve? Manually reviewing long talking-head or narration footage to remove redundant speech and keep valuable segments is slow and repetitive. This Skill automates the rough cut by analyzing the video's audio track, recognizing speech with timestamps, and producing a structured JSON edit decision list for downstream timeline generation. ## Core Features & Use Cases - ASR-Based Speech Analysis: Runs an ASR node to transcribe narration and label word-level timestamps from the input video. - Automated Rough Cutting: Uses a speech_rough_cut node to split the video into kept and removed segments, with support for custom user requests via the user_request parameter. - Timeline and Render Output: Generates a timeline with is_speech_rough_cut=True, recommends caption styles, and renders while preserving the original audio. - Use Case: A creator uploads a 30-minute talking-head recording and asks the assistant to cut out pauses, filler words, and repeated takes, receiving a ready-to-render rough cut timeline. ## Quick Start Perform a narration rough cut on the attached video, removing redundant speech and generating a timeline that keeps the original audio.
