---
name: musicvideo-track
description: "Prepare the person's track for an AI music video — lyrics with word timestamps (ElevenLabs Scribe v2 on fal), style and mood (fal-ai/audio-understanding), the beat grid measured from the audio (spectral flux autocorrelation; a model's BPM can be wrong), and parts of ≤15 s cut on a beat inside a vocal pause. Part of the musicvideo reference set."
---

These creative references describe source experiments with external tools. Use your own supported account and tools; this download installs no automation. Settings and timings are experimental observations, not current FlashBoards availability. Use only media and voices you have permission to use.

# musicvideo-track · lyrics, style, beats, parts

Work on the person's own `{track}`. Compute everything fresh for it; never reuse another run's lyrics, beats or part times.

## Steps

1. **Lyrics** — `fal-ai/elevenlabs/speech-to-text/scribe-v2` with `tag_audio_events: true` → words with `start`/`end` (the word is in `text`).
2. **Style** — `fal-ai/audio-understanding`: genre, voice, mood, structure. Use it for words, not numbers: on the demo track it said 150 BPM; the audio has **132.5 BPM** — always measure.
3. **Beat grid** — measure it: onset strength (spectral flux) → autocorrelation → tempo and phase. Store `beats.json`: `beat` = seconds per beat, `phase` = time of the first beat.
4. **Parts** — H3 takes ≤ 15 s. From the current start look 11–15 s ahead, find the **longest gap between sung words** in that range, cut on the beat nearest to it; repeat from the cut to the end of the track. Generation length = part rounded up to whole seconds.
5. **Audio per part** — cut `{track}` for each part (mp3) and upload; it is **Audio 1** of that part.

## Example run (reference only — never reuse its files, prompts, faces or frames for a person)

Demo track «3upd (Cover)», 88.8 s: `beat` = 0.4528 s, `phase` = 0.035 s for 132.5 BPM (the style model had said 150). Parts: 0–11.81, 11.81–26.30, 26.30–41.24, 41.24–52.56, 52.56–64.78, 64.78–79.27, 79.27–88.80 (7 parts).