Skip to main contentZark processes audio and video files to understand spoken content, visual moments, and time-based information.
When media files are processed, Zark:
- Identifies media type and extracts audio
- Generates transcript with timestamps
- Analyzes content for structure, intent, and important moments
- Indexes everything for natural language search
Processing runs automatically. No manual configuration is required.
Each media file produces searchable data:
Transcripts
Every spoken word is captured with timestamps. Speaker attribution is included when possible.
Key Moments
Zark detects important moments:
- Decisions
- Action items
- Key statements
- Highlights
- Introductions and conclusions
Topics and Entities
People, companies, products, and themes are identified and indexed.
Visual Context (Video Only)
For video files, Zark detects:
- Slides and on-screen text
- Charts and diagrams
- Product demos
- Scene changes
| Media Type | Formats |
|---|
| Video Files | MP4, MOV, AVI, MKV, WebM |
| Audio Files | MP3, WAV, M4A, FLAC, OGG |
- Finding what was said by meaning, not exact words
- Locating moments by topic or event
- Identifying speakers and their statements
- Understanding visual content alongside audio
See How Indexing Works for how media is indexed.
What Becomes Searchable
After processing, media becomes searchable by:
- Topics discussed
- People mentioned
- Events and decisions
- Specific moments and timestamps
Search automatically combines semantic understanding, entity detection, moment classification, and exact phrase matching. See Search & Discovery for how search works. 
