Generate YouTube shorts from prompts with OpenAI, ElevenLabs and Pexels

by Davide Boizza ✓ · 5,187 views · created 2026-06-16 · Content Creation, Multimodal AI

What this workflow does

This AI workflow has 18 nodes and runs when someone submits an n8n form (“Short-form vertical video content”). Along the way it connects YouTube and OpenAI Chat Model. It also calls api.elevenlabs.io and queue.fal.run over HTTP.

How it runs

Trigger: When someone submits an n8n form (“Short-form vertical video content”).

1. Set vars (Edit Fields (Set)): sets or maps fields
2. Generate voiceover (AI Agent): runs an AI agent, with model: OpenAI Chat Model; output parser: Structured Output Parser
3. Voiceover vars (Edit Fields (Set)): sets or maps fields
4. Text-to-Speech (HTTP Request): calls api.elevenlabs.io (POST)
5. Pexels (HTTP Request): calls an HTTP API
6. Upload audio (FTP): upload
7. Merge: merges the data from its input branches
8. Set Audio Url (Edit Fields (Set)): sets 1 field: audio_url
9. Merge Audio with Video (HTTP Request): calls queue.fal.run (POST)
10. Wait: waits 30 seconds

…plus 5 more steps. See the full node list below.

Services it uses

YouTubeOpenAI Chat Model

HTTP calls to: api.elevenlabs.io, queue.fal.run

Before you upgrade to n8n 3.0

Nodes used (11)

• HTTP Requestn8n-nodes-base.httpRequest
• Mergen8n-nodes-base.merge
• Ifn8n-nodes-base.if
• Waitn8n-nodes-base.wait
• Edit Fields (Set)n8n-nodes-base.set
• YouTuben8n-nodes-base.youTube
• OpenAI Chat Model@n8n/n8n-nodes-langchain.lmChatOpenAi
• n8n Form Triggern8n-nodes-base.formTrigger
• Structured Output Parser@n8n/n8n-nodes-langchain.outputParserStructured
• AI Agent@n8n/n8n-nodes-langchain.agent
• FTPn8n-nodes-base.ftp

From the author

Quick overview This workflow takes a prompt from an n8n form, uses OpenAI to generate a YouTube Shorts script, then creates an ElevenLabs voiceover, pulls matching portrait stock video from Pexels, merges audio and video via fal.run’s FFmpeg API, and uploads…

View the original on n8n.io → Check your own copy