Generate AI-powered notes from videos (document, outline, or graphic-text formats)
Install
Documentation
AI Video Notes
Generate structured notes from video URLs using Baidu AI. Supports three note formats.
Workflow
1. Create Task: Submit video URL → get task ID
2. Poll Status: Query task every 3-5 seconds until completion
3. Get Results: Retrieve generated notes when status = 10002
Status Codes
| Code | Status | Action |
|-------|---------|---------|
| 10000 | In Progress | Continue polling |
| 10002 | Completed | Return results |
| Other | Failed | Show error |
Note Types
| Type | Description |
|------|-------------|
| 1 | Document notes |
| 2 | Outline notes |
| 3 | Graphic-text notes |
APIs
Create Task
Endpoint:POST /v2/tools/ai_note/task_create
Parameters:
- -
video_url(required): Public video URL
python3 scripts/ai_notes_task_create.py 'https://example.com/video.mp4'
Response:
{
"task_id": "uuid-string"
}
Query Task
Endpoint:GET /v2/tools/ai_note/query
Parameters:
- -
task_id(required): Task ID from create endpoint
python3 scripts/ai_notes_task_query.py "task-id-here"
Response (Completed):
{
"status": 10002,
"notes": [
{
"tpl_no": "1",
"contents: ["Note content..."]
}
]
}
Polling Strategy
Option 1: Manual Polling
1. Create task → store task_id
2. Query every 3-5 seconds:
python3 scripts/ai_notes_task_query.py <task_id>
3. Show progress updates:
- Status 10000: Processing...
- Status 10002: Completed
4. Stop after 30-60 seconds (video length dependent)
Option 2: Auto Polling (Recommended)
Use the polling script for automatic status updates:
python3 scripts/ai_notes_poll.py <task_id> [max_attempts] [interval_seconds]
Examples:
Default: 20 attempts, 3-second intervals
python3 scripts/ai_notes_poll.py "task-id-here"
Custom: 30 attempts, 5-second intervals
python3 scripts/ai_notes_poll.py "task-id-here" 30 5
Output:
- -Shows real-time progress:
[1/20] Processing... 25% - -Auto-stops when complete
- -Returns formatted notes with type labels
Error Handling
- -Invalid URL: "Video URL not accessible"
- -Processing error: "Failed to parse video"
- -Timeout: "Video too long, try again later"
Launch an agent with ai notes of video on Termo.