AI Import Setup Guide
Overview
The AI import feature allows you to automatically transcribe and generate content from YouTube videos and other media URLs using AssemblyAI for transcription and OpenAI for content generation.
Setup Instructions
-
Copy the environment file
cp .env.example .env
-
Get your API keys:
-
AssemblyAI API Key (Required)
-
OpenAI API Key (Required)
-
YouTube API Key (Optional, for better metadata)
- Go to https://console.cloud.google.com
- Create a new project or select existing
- Enable YouTube Data API v3
- Create credentials (API Key)
- Add to
.env: YOUTUBE_API_KEY=your_key_here
-
Restart your dev server
npm run dev
Usage
- Navigate to any media collection (Talks, Hosted Podcasts, Guest Podcasts)
- Click “Import from URL”
- Paste a YouTube URL
- Select processing options
- Click “Process Media”
- Review and edit the generated content
- Save as draft or publish
Supported URLs
- YouTube: Direct video URLs (youtube.com/watch?v=…, youtu.be/…)
- Downloads audio with yt-dlp and transcribes with AssemblyAI
- Includes speaker diarization to identify different speakers
- Requires yt-dlp to be installed:
brew install yt-dlp (macOS) or pip install yt-dlp
- Vimeo: Video URLs (vimeo.com/…)
- Direct audio/video files: MP3, MP4, WAV, etc. (must be publicly accessible)
API Status
The API integration is now fully enabled with real AssemblyAI and OpenAI calls. The system will:
- Submit YouTube URLs directly to AssemblyAI
- Poll for transcription completion
- Generate content using GPT-4
- Include speaker diarization and auto-generated chapters
Cost Estimates
- AssemblyAI: ~$0.65 per hour of audio ($0.0108 per minute)
- OpenAI GPT-4: ~$0.03 per 1K tokens (typical video: $0.10-0.30)
- Total per video: ~$0.50-1.00 depending on length