Self-Hosted n8n Workflow: Gemini/Claude Shorts w/ Edge TTS & FFmpeg
Self-Hosted n8n Workflow: Gemini/Claude Shorts w/ Edge TTS & FFmpeg
Regular price
£28.99
Regular price
£28.99
Sale price
Unit price
/
per
⬇
Instant Digital Download
∞
Unlimited Downloads
★
Lifetime Access in Your Account
Couldn't load pickup availability
🔥
128+ Sold
Popular with n8n builders
⚡
23 people viewing
High interest right now
✅
9 added today
Fast-moving digital product
Self-Hosted n8n Workflow: Gemini/Claude Shorts w/ Edge TTS & FFmpeg
Regular price
£28.99
Regular price
£28.99
Sale price
Unit price
/
per
Turn an n8n topic into a ready-to-post vertical short—self-hosted
This self-hosted n8n workflow takes a topic from an n8n Form Trigger, generates a multi-scene short using Google Gemini or Anthropic Claude, creates synchronized voiceover with Edge TTS, downloads scene visuals from Pollinations, and renders a final vertical MP4 using FFmpeg (with karaoke-style burned captions).
What this workflow does
- Collect inputs: Receives topic, language, scene count, and LLM choice from an n8n Form Trigger.
- Voice + pacing logic: Maps the selected language to an Edge TTS voice and calculates a per-scene word limit to keep the total video under 60 seconds.
- Generate the script: Produces a JSON script containing voiceover text plus English image prompts per scene using either Google Gemini or Anthropic Claude.
-
Scene-by-scene rendering:
- Splits the script into scenes and generates an MP3 voiceover per scene with Edge TTS.
- Estimates caption timing word-by-word using ffprobe/FFmpeg silence detection.
- Downloads a 1080×1920 image per scene from Pollinations based on the scene prompt.
- Renders each scene locally with FFmpeg: applies a slow zoom effect and burns karaoke-style ASS subtitles synced to the audio.
- Finalize output: Concatenates all scene segments into a single vertical MP4, normalizes audio loudness, and returns the final video as the workflow output.
Use cases
- Generate TikTok/Shorts-style vertical videos from form submissions without third-party video tools.
- Automate “script → voice → captions → render” for SaaS marketing, creator ops, or rapid testing of content ideas.
- Keep data and rendering self-hosted while switching between Gemini and Claude.
Technical details
- LLMs: Google Gemini or Anthropic Claude
- Voice: Microsoft Edge TTS
- Images: Pollinations (1080×1920 per scene)
- Video & captions: FFmpeg + ffprobe (libass for burned ASS subtitles)
- n8n nodes referenced: code, merge, switch, split out, aggregate, sticky note
-
Self-host setup: Start n8n with
NODES_EXCLUDE="[]"andN8N_RESTRICT_FILE_ACCESS_TO="/tmp"(macOS:"/tmp;/private/tmp") and enable local Execute Command + file access.
