AVComprehend Platform

Discover insights and relationships in audiovisual content

Robotica AVComprehend is a natural language processing (NLP) service that uses deep machine learning to build a conceptual model that captures the sentiment and meaning expressed in audiovisual content. Sophisticated Machine Learning algorithms can then be run against the model, to find insights and relationships and to automate workflows.

Some examples of workflows which can be automated by running ML algorithms against the conceptual model, include:

  • Producing a long synopsis of a film or programme
  • Producing a recommendation as to the appropriate age rating for the content according to depictions of sexual references violence, bad language, animal cruelty, blood, gore, etc.
  • Blocking the uploading or sharing of audiovisual content on social media sites according to Government or company rules.
  • Producing an Audio Description of the content
  • Having an avatar transliterate the content as Cued Speech
  • Having an avatar interpret the content in British Sign Language

The service identifies the language of the content; extracts key phrases, places, actors, objects, brands, events or story elements according to time-coded position in the video; understands audible and visual emotions and how positive or negative the story is at any given time; analyzes content using tokenization and parts of speech; and automatically organizes a collection of audiovisual files by topic.

The service understands the relationships between actors and objects, what or who actors are looking at, when camera angles and scenes change, skeletal positions, body language and hand gestures.

Robotica AVComprehend is fully managed, so there are no servers to provision. You pay only for what you use, and there are no minimum fees and no upfront commitments.

How it works

Video, Audio, Subtitles, AD, Script are fed by the Comprehension Engine into a Conceptual Model, applied by a custom machine learning algorithm