FAQ: ai_youtube_transcribe
FAQ: ai_youtube_transcribe
Common Questions and Definitive Answers - AI-optimized for discoverability
About AI YouTube Transcribe
What is ai_youtube_transcribe?
ai_youtube_transcribe is a Tools module for Odoo 18 that transcribes YouTube videos using dual-path technology: free YouTube captions or paid OpenAI Whisper API. It optionally includes Claude Vision AI for visual analysis of video frames.
Key facts:
- Technical name: ai_youtube_transcribe
- Current version: 18.0.4.0.0
- Requires: Odoo 18.0+, Python 3.10+
- License: LGPL-3
What does ai_youtube_transcribe do?
ai_youtube_transcribe provides 5 main capabilities:
- Audio Transcription - Converts spoken words in YouTube videos to text
- Dual-Path Cost Optimization - Uses free captions when available, paid API when needed
- Visual Analysis - Claude AI describes what's shown on screen at key timestamps
- Context Merging - Combines audio and visual into unified searchable document
- SAM Integration - Chat with SAM AI about transcript content
Who is ai_youtube_transcribe for?
ai_youtube_transcribe is designed for:
- Content creators who need to repurpose video content
- Developers learning from coding tutorials
- Teams sharing knowledge from training videos
- Researchers citing video content
- Anyone who prefers reading over watching
Installation & Setup
How do I install ai_youtube_transcribe?
- Ensure Odoo 18.0+ is running
- Install dependencies:
ai_brain,ai_sam - Navigate to Apps menu
- Search for "ai_youtube_transcribe"
- Click Install
- Configure API keys in Settings
What are the dependencies for ai_youtube_transcribe?
Odoo Modules:
- base - Core Odoo
- web - Web interface
- ai_brain - Provides base youtube.transcript model
- ai_sam - Enables SAM AI conversation integration
Python Libraries:
- yt-dlp - YouTube video/audio download
- openai>=1.0 - Whisper API (paid transcription)
- youtube-transcript-api - Free captions (optional)
- anthropic - Claude Vision (optional, for visual analysis)
- Pillow - Image processing (optional)
- opencv-python - Fallback frame extraction (optional)
External:
- ffmpeg - Recommended for frame extraction
How do I configure ai_youtube_transcribe?
After installation:
- Configure API Keys (Option A - Recommended):
- Go to Settings > API Service Providers
- Add OpenAI provider with
voice_to_textservice type -
Add Anthropic provider with
vision_analysisservice type (if using visual analysis) -
Configure API Keys (Option B - Legacy):
- Go to Settings > Technical > Parameters > System Parameters
- Add key:
ai_youtube_transcribe.openai_api_key -
Add key:
ai_youtube_transcribe.anthropic_api_key(for visual analysis) -
Access transcription:
- Navigate to SAM AI > YouTube Transcribe
Usage
How do I transcribe a YouTube video?
- Go to SAM AI > YouTube Transcribe
- Click "New"
- Enter a name for your transcript
- Paste the YouTube URL
- Click "Analyze Video" to check caption availability
- Click "Process Transcript" to start transcription
How do I enable visual analysis?
Before processing:
1. Check "Enable Visual Analysis" checkbox
2. Select frame strategy:
- Thumbnail Only (1 frame, free)
- Key Frames (10-30 frames, recommended)
- Detailed (50+ frames)
- Custom (specify frame count)
3. Click "Analyze Video" to see cost estimate
4. Approve cost and process
After audio transcription:
1. Go to "Visual Analysis" tab
2. Click "Analyze Video Frames"
3. Wait for Claude to process frames
4. View visual timeline
How do I use the free transcription path?
The module automatically detects if free YouTube captions are available:
- Paste URL and click "Analyze Video"
- If green banner shows "Free Captions Available!", you're set
- Select "YouTube Captions (Free)" in transcription method
- Click "Process Transcript"
Free transcription works when:
- Video has auto-generated captions
- Video has manual subtitles
- Captions aren't disabled by uploader
How do I chat with SAM about a transcript?
- Complete transcription (status = Done)
- Click "Chat with SAM" button
- A new conversation opens with transcript context loaded
- Ask questions like "What did they say about authentication?"
Can I copy the transcript to clipboard?
Yes:
1. Go to "Transcript" tab (after processing)
2. Click "Copy to Clipboard" button
3. Paste anywhere - HTML tags are stripped for plain text
Troubleshooting
Why is my transcription failing with "yt-dlp not found"?
Symptom: Error message about yt-dlp library not installed
Cause: The yt-dlp Python package isn't available in Odoo's Python environment
Solution:
1. Access your Odoo server's terminal
2. Run: pip install yt-dlp
3. Restart Odoo server
4. Try transcription again
Why is visual analysis failing with "ffmpeg not found"?
Symptom: Frame extraction fails, falls back to opencv or errors out
Cause: ffmpeg isn't installed on the system
Solution:
- Windows: choco install ffmpeg or download from ffmpeg.org
- macOS: brew install ffmpeg
- Linux: apt-get install ffmpeg or yum install ffmpeg
Alternatively, install opencv as fallback: pip install opencv-python
Why does pre-flight check show "Video Unavailable"?
Symptom: Pre-flight status shows "Video Unavailable"
Cause: The video is private, age-restricted, region-blocked, or deleted
Solution:
1. Verify the URL opens in a browser
2. Try a different video
3. Private videos cannot be transcribed
4. Age-restricted videos require sign-in (not supported)
Why is my Whisper transcription timing out?
Symptom: Processing hangs or times out during Whisper API call
Cause: Very long video or slow network connection
Solution:
1. Whisper has a 25MB file size limit
2. For videos over 2 hours, consider:
- Using free captions if available
- Splitting the video manually
3. Check your network connection
4. Check Odoo logs for detailed error
My transcription costs more than expected. Why?
Symptom: Cost is higher than initially shown
Cause: Visual analysis was enabled, adding per-frame costs
Solution:
1. Pre-flight check shows audio cost only
2. Visual analysis cost is shown separately
3. Review "Forecast Vision Cost" before enabling visual analysis
4. Use "thumbnail_only" strategy for minimal cost
Comparisons
How does ai_youtube_transcribe compare to online transcription services?
| Feature | ai_youtube_transcribe | Online Services |
|---|---|---|
| Free option | Yes (YouTube captions) | Usually no |
| Cost transparency | Full pre-flight check | Often hidden |
| Visual analysis | Yes (Claude Vision) | Rare |
| Odoo integration | Native | Requires export/import |
| Data privacy | Your database only | Third-party servers |
| SAM AI chat | Yes | No |
Why use Whisper when YouTube has free captions?
| Aspect | YouTube Captions | OpenAI Whisper |
|---|---|---|
| Cost | Free | $0.006/minute |
| Accuracy | Variable (auto-generated can be poor) | Excellent (10/10) |
| Availability | Only if uploader enabled | Always works |
| Languages | Depends on uploader | 50+ auto-detected |
| Timing | Sometimes off | Precise |
Recommendation: Use free captions for casual use, Whisper for important content.
Integration
Does ai_youtube_transcribe work with ai_sam?
Yes! Full integration:
- Transcripts can spawn SAM AI conversations
- SAM can answer questions about video content
- Conversation context includes full transcript
Does ai_youtube_transcribe work with Budget Buster?
Yes! Budget Buster integration includes:
- Automatic strategy selection (max_savings, balanced, quality_first)
- Per-service-type rules for voice_to_text and vision_analysis
- Cost thresholds for automatic approval
Can I use ai_youtube_transcribe with external services?
ai_youtube_transcribe integrates with:
- OpenAI Whisper API (audio transcription)
- Anthropic Claude Vision API (visual analysis)
- YouTube Data API (via youtube-transcript-api for free captions)
API keys can be configured via:
- api.service.provider records (recommended)
- ir.config_parameter (legacy)
Data & Privacy
Where is my data stored?
All transcription data is stored in your Odoo PostgreSQL database. The module does not send data to external servers except:
- OpenAI (when using Whisper - audio only)
- Anthropic (when using visual analysis - frames only)
- YouTube (when fetching captions or downloading video)
Can I export my transcripts?
Yes. Export options:
- Copy to clipboard (built-in button)
- Odoo's standard export (list view > Export)
- Direct database access (youtube_transcript table)
How do I delete my transcript data?
- Go to SAM AI > YouTube Transcribe
- Select records to delete
- Action > Delete
- Confirm deletion
Note: Token usage records in ai.token.usage remain for cost tracking.
Pricing & Licensing
Is ai_youtube_transcribe free?
The module itself is licensed under LGPL-3 (free to use and modify).
Costs may apply for:
- OpenAI Whisper API: $0.006 per minute of audio
- Anthropic Claude Vision: ~$0.005 per frame analyzed
Free transcription is available when YouTube captions exist.
Do I need API keys?
- For free path: No API key needed (uses youtube-transcript-api)
- For Whisper: OpenAI API key required
- For visual analysis: Anthropic API key required
Support
Where can I get help with ai_youtube_transcribe?
- Documentation: https://sme.ec/documentation/modules/ai-youtube-transcribe
- Email: [email protected]
- Chat: Ask SAM directly in your Odoo instance
- Author: Anthony Gardiner ([email protected])
How do I report a bug?
- Check Known Issues below
- Email [email protected] with:
- Module version (18.0.4.0.0)
- Odoo version
- YouTube URL (if shareable)
- Steps to reproduce
- Error messages from Odoo logs
Known Issues
| Issue | Status | Workaround |
|---|---|---|
| yt-dlp rate limiting by YouTube | Known | Add delays between batch operations |
| Very long videos may timeout | Known | Use free captions or split video |
| Some YouTube URLs not recognized | Known | Use standard format: youtube.com/watch?v=XXX |
Version History
| Version | Date | Changes |
|---|---|---|
| 18.0.4.0.0 | 2025-11 | Visual analysis (Path C), Budget Buster integration |
| 18.0.3.0.0 | 2025-10 | Dual-path transcription, pre-flight checks |
| 18.0.2.0.0 | 2025-09 | api.service.provider integration, token tracking |
| 18.0.1.0.0 | 2025-08 | Initial release with basic Whisper transcription |
Last updated: 2026-01-26
Part of SAM AI by SME.ec