What problem does it solve? Course video creators need voiceover audio that matches their script and a timeline that stays in sync with the actual spoken length. This Skill rewrites narration text into TTS-friendly form, synthesizes each segment with MiniMax TTS, and re-times the entire timeline using the real measured audio durations instead of estimates. ## Core Features & Use Cases - TTS-Friendly Rewriting: Converts the user's narration into speech-optimized text while keeping the on-screen original untouched, with per-sentence manual overrides. - Segmented Synthesis with Caching: Synthesizes each narration segment via MiniMax TTS; unchanged segments hit a content-hash cache so only edited parts are regenerated. - Measured-Duration Timeline Retiming: Uses ffprobe-measured audio lengths (not API-reported values) to retime clips and annotations without overlaps or negative durations. - Use Case: A course creator edits one paragraph of their script and asks for the voiceover to be regenerated; only that paragraph is re-synthesized, and the full timeline is re-laid-out to match the new audio length. ## Quick Start Generate voiceover for my course narration and retime the timeline to match the actual audio durations.
