Video Breakdown and Original-Timeline Transcripts

What it does

Prepare a video for editing by splitting it into manageable sections and collecting audio, review images and transcripts. Original-timeline timestamps help you find each spoken passage in the source video.

• Export audio and optional video clips at a chosen interval
• Create representative images for each section
• Generate transcripts mapped to the original timeline
• Save text, subtitle files, review reports and an editing reference table
• Preview output counts, protect existing files and resume matching work

Local videos are supported. Optional YouTube intake is limited to material you own or are authorized to use and whose source service permits downloading. This is not an unrestricted video downloader.

How to use

Load the skill in a supported AI agent and specify your source and desired interval. Confirm permission, prepare the source, review the output plan, split, transcribe and verify the results.

Before you start

A separate runtime, external tools and a speech-recognition model are required. Their names and setup instructions are included in the README and workflow guide. Executables and models are not bundled. Initial model setup requires a download; transcription can then run locally using the prepared model. CPU operation has been tested, and a GPU is not required. Processing time depends on your computer, model and video length.

Caveats

Use your own videos or properly authorized material. No unauthorized access to private or restricted videos or bypassing protection is supported. Stop when the source rejects a download.

Splitting and transcription were tested end to end on original spoken video on Windows. The downloader was exercised with a controlled test source; real YouTube retrieval, macOS and GPU operation remain unverified.

Transcripts can contain incorrect names or numbers, hallucinations during silence and missing speech around split boundaries. Accuracy is not guaranteed. Compare results with the original audio before publishing.