n8n Workflow: Captioned Video Highlights w/ Zvid & Webhook

Turn any MP4 into captioned vertical Shorts—automatically—using n8n, Zvid, ElevenLabs Scribe, and a webhook

This n8n workflow takes a direct MP4 URL (via POST webhook or manual test), transcribes the opening with ElevenLabs Scribe, selects highlight ranges with OpenRouter, and then builds Zvid projects to render captioned 1080x1920 Shorts. You’ll get back the final clip URLs (or editor drafts) in the webhook response.

What this workflow does

• Receives input: Accepts a direct video URL and optional title/layout via webhook, or runs with a built-in sample video for testing.
• Extracts timestamps: Downloads a byte-range sample and sends it to ElevenLabs Speech-to-Text (Scribe) to obtain word-level timestamps.
• Finds highlights: Formats the transcript into an indexed prompt and uses OpenRouter (Gemini) to choose non-overlapping highlight ranges and short on-screen titles.
• Trims and captions: Converts selected word ranges into exact start/end timestamps, trims clips to configured min/max duration, and prepares caption timing.
• Creates Zvid projects: Generates per-clip Zvid render project JSON (vertical layout, captions, branding, progress bar) and validates with Zvid’s render validator.
• Dry-run or render: With dryRun, saves Zvid editor drafts and returns links; otherwise renders, polls for completion, and returns final rendered clip URLs.

Use cases

• SaaS operators: auto-generate captioned Shorts from user-submitted MP4s on upload.
• Marketing teams: produce multiple highlight clips per source video for Reels/TikTok with consistent caption styling.
• Automation engineers: a reusable webhook-based pipeline for video highlight generation with robust timestamping.

Technical details

• Nodes / logic: if, set, code, wait, webhook, and Zvid integration via @zvid/n8n-nodes-zvid.
• Integrations: ElevenLabs Scribe (speech-to-text with word timestamps) and OpenRouter (model-based highlight selection).
• Output: captioned Zvid renders at 1080x1920, returned as clip links in the webhook response.