Song-to-Video Script Builder · सार्वजनिक Open Skill
Song-to-Video Script Builder
Song-to-Video Script Builder Digen पर एक सार्वजनिक Open Skill है। खोलें और जानें कि यह वर्कफ़्लो क्या बनाता है।
Music
8 1 5
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/1716469/2026-10-08/5pczjkgg.jpg
--- allowed-tools: - transcribe_audio - create_document - update_document - get_document - generate_image - generate_tts --- # Song-to-Video Script Builder ## Reference files - `references/beat-map-template.md`: Subject-matter beat-map template with a worked example. Read it in Step 4 before building the beat-level references. ## When to use - The user uploads a completed song (audio file or link) and wants the **entire song transcribed into a timestamped video script** that closely follows the story the song is telling. - The user wants a **formatted text-file script ready for Video Director**: every scene tied to exact timestamps, vocal lines timestamped for lip-sync, and on-screen actions/visuals derived from the lyrics. - The user wants a script with **beat-level references** — every scene's visuals closely tied to the subject matter of what is being said in the song, so the script reflects what the song is all about. - The user wants to revise or re-transcribe an existing script (fix lyric text, re-time a line, expand a scene, adjust the story interpretation). - The user wants **storyboard images** and/or **voiceover** generated from the script to pre-visualize the video. ## Steps ### 1. Ingest the song and transcribe the full length - Take the uploaded audio (file or URL) and run `transcribe_audio(audio_url=...)` (channel `asr`) to get the complete transcript plus timestamped segments (`dft` / `srt`). - Cover the **entire song, start to finish** — every vocal line, and instrumental-only stretches marked as such. Never truncate. ### 2. Build the timestamped vocal timeline - From the timestamped segments, list every vocal line with its exact start time, end time, and lyric text (e.g. `[0:14–0:18] "…"`). - Flag any uncertain or overlapping lines and ask the user to confirm the lyric text before finalizing. - These timestamps are the script's backbone — Video Director and lip-sync both depend on them. ### 3. Map the story and structure the script - Read the full lyrics as a narrative: what story is the song telling, who is it about, what happens, and where do the mood and energy turn? - Divide the song into sections that follow the song's own structure (intro / verse / chorus / bridge / outro, or its natural beats). For each section, note: time range, what is happening in the story, and the mood/energy. - The script must **closely adhere to the story the song tells** — every visual beat follows the narrative implied by the words. ### 4. Create beat-level references (tie the script to the subject matter) - Read `references/beat-map-template.md` and use its template to structure the beat map. - Go through the song beat by beat (each vocal line / musical segment) and pin down exactly what **subject matter** is being said at that moment — the concrete topic, image, or idea the lyrics express. - For each beat, write a **subject-matter reference**: what is being said, what it means in the story, and the visual that should reflect it. - Ensure every scene's visual direction is **closely tied to the subject matter of the lyrics being said at that moment** — the on-screen action, setting, and imagery should mirror the words, not drift into unrelated visuals. - Use these subject-matter references to drive the scene-by-scene script below, so the script closely reflects what the song is all about. ### 5. Write the timestamped script and save it as a formatted text file - Produce a scene-by-scene script covering the **whole song** from 0:00 to the final timestamp. For each scene include: - **Time range** (e.g. `0:00–0:12`) - **Section + mood/energy** - **Beat-level references**: the subject matter being said in this scene and the visual that reflects it - **Visual direction**: on-screen action, setting, camera, and lighting — derived from what the vocals are saying at that moment, telling the story - **Vocal lines with exact timestamps** (for lip-sync) - Format every scene in the output file using the **sc