Intelligent metadata extraction by understanding video, text, audio and images.

No training or setup fees
Highest accuracy powered by a self-learning knowledge graph


Our multi-modal processing looks at all components of a video to provide accuracy and relevance:

  • Visual detection models to capture faces, landmarks, objects, logos and more
  • AI-assisted video editing and breakpoints
  • Shot detection and topic segmentation
  • Optical character recognition (OCR) for on-screen text
  • Video sentiment analysis
  • Speech to text combined with our Knowledge Graph for deeper understanding


  • Process URLs, documents or raw text
  • Natural language processing and NER for optimal entity recognition and linking
  • Multi-language support: English, Spanish, Portuguese, French, German, Dutch, Catalan, Greek, Italian and Hindi NEW. New languages continually added
  • Native language support means no translation required, reducing cost


  • Multi-language speech to text gives you automated transcriptions and extracts key information
  • Corrects common mistakes through contextual analysis and disambiguation from our Knowledge Graph
  • Enables time-coded searching and topic segmentation


  • The same powerful computer vision used for video processing, at the image level
  • Deeper contextual tagging thanks to our Knowledge Graph
  • Make your entire raw image archive searchable with automated indexing

Why Metaphor?

Metaphor metadata engine is powered by the three unique pillars of Vilynx technology.

Knowledge graph

A language-agnostic entity base that grows and evolves constantly, adding millions of concepts and billions of relationships monthly. This means fewer errors, deeper understanding and better metadata.


Automatically learns and adapts to your content, meaning no manual training or setup fees.


Evaluates all components of content to determine the most relevant tags from our knowledge graph, using computer vision, optical character recognition, speech to text and natural language processing.

How our Knowledge Graph gives you the best tagging

Our Knowledge Graph allows us to provide deeper context and better disambiguation to understand your data. What that means for you is tags that are more detailed, relevant and useful.

Take this first text example. “Barcelona” could refer to multiple things, most notably the Spanish city. Same with “Messi.” However, using our Knowledge Graph, we’re able to disambiguate these terms to find the correct concepts they are referring to; in this case the soccer club (not the city) and its star player Lionel Messi.

Furthermore, we can extrapolate that this entire text is related to soccer, even though there is no mention of it. This kind of contextual understanding is another benefit our Knowledge Graph provides.

Without a Knowledge Graph, you will get incorrect, generic and irrelevant tags that are not very useful for your business. You’re probably familiar with this if you’ve tried other metadata tagging services.