Auto-Create Podcast from YouTube Video Transcript using Dumpling AI and GPT-4o

Introduction

Repurposing video content into audio can be time-consuming—but this automation changes everything. With just a YouTube link and a bit of AI magic, this workflow automatically:

1. Monitors a YouTube RSS feed for new uploads
   
2. Scrapes the full transcript using Dumpling AI
   
3. Cleans and summarizes it using GPT-4o
   
4. Saves the podcast script and metadata directly into Airtable
   

This setup is perfect for content creators, podcast editors, or media agencies who want to scale their content library from a single video.

Workflow Overview

The workflow consists of four main steps:

1. Watch for New YouTube Video via RSS

• Node Type: RSS Feed Read Trigger
  
• What it Does: This trigger monitors a YouTube RSS feed (from rss.app) for any new video uploads. Once it detects a new video, the workflow begins.
  
• Why It Matters: It automates detection so you don’t need to manually paste URLs. As soon as a video drops, it gets processed.
  

Feed URL Example:
https://rss.app/feeds/Vw076Uzh7bIinpci.xml

2. Get Transcript from YouTube Video using Dumpling AI

• Node Type: HTTP Request
  
• Method: POST
  
• URL: https://app.dumplingai.com/api/v1/get-youtube-transcript
  
• Headers: Authenticated using httpHeaderAuth
  
• Payload:

{

  “videoUrl”: “{{ $json.link }}”,

  “preferredLanguage”: “en”

}

• What it Does: This sends the YouTube video link to Dumpling AI’s endpoint, which scrapes and returns the full transcript.
  
• Why It Matters: You get a clean transcript without needing a manual download or YouTube API key.
  

Pro Tip: You can set fallback language preferences or error handling here if a transcript is unavailable.

3. Transform Transcript into Podcast Script using GPT-4o

• Node Type: OpenAI Chat Model
  
• Model: chatgpt-4o-latest
  
• Prompt Logic:
  

The system prompt instructs GPT-4o to:

• Remove filler words (like “um”, “you know”, “actually”, etc.)
  
• Label each speaker clearly (e.g., Speaker 1:)
  
• Combine all dialogue into one cleaned block
  
• Summarize the conversation in 2–4 sentences
  
• Generate a podcast-appropriate title
  

Prompt Snippet:

{

  “title”: “Relevant podcast title here”,

  “cleaned_transcript”: “Speaker 1: Text. Speaker 2: Text.”,

  “summary”: “Short summary of the conversation.”

}

• Why It Matters: You go from a raw transcript to a polished podcast script and summary—automatically.

4. Save Podcast Script and Metadata to Airtable

• Node Type: Airtable (Create Record)
  
• Table Name: podcast
  
• Base: Testing n8n
  
• Mapped Fields:
  
  • Title: The generated podcast title
    
  • podcast transcript: The cleaned transcript output
    
  • summary: The 2–4 sentence summary generated by GPT-4o
    
• Why It Matters: All outputs are stored for easy access, editing, or further automation (like publishing or scheduling).

Optional Enhancements

• Audio Generator: Add a TTS (Text-to-Speech) node to convert the transcript into MP3 format.
  
• Publishing: Connect to WordPress or Zapier to auto-publish new episodes.
  
• Newsletter: Email summaries directly using the Gmail node for instant distribution.

Conclusion

This automation is a complete podcast production assistant. It starts with a YouTube video and ends with a structured, high-quality podcast script—ready to publish. By combining Dumpling AI and GPT-4o, you eliminate all the grunt work of transcription, editing, and formatting.

Whether you’re running a content repurposing agency or just want to keep your audience engaged across platforms, this setup saves hours and keeps quality high.