Profuz Digital showcased the latest edition of its LAPIS media asset management system at the NAB Show last week, adding AI technology like semantic search, speech to text, translation and face and object recognition.
The updates include:
- Semantic Search: Enables search by meaning across all text-based data, in addition to obvious keyword or phrase searches.
- Speech-to-Text: Converts audio and video into timed text. Depending on the engine, various languages are supported, including for direct audio translation.
- Speaker Identification: Detects and labels individual speakers within audio and video content.
- Text-to-Speech: Supports playback of both static and timed text.
- Text Translation: Supports translation of plain and timed text files, with built-in management for engines that do not natively support timed text.
- Object Recognition: Automatically detects objects within images and video content, enabling sub-image classification and advanced search.
- Face Recognition: In LAPIS you can add identity photos that can be used to recognise people in images and videos. Unknown faces can be classified manually. This data is also used in search results.
- Optical Character Recognition (OCR): Extracts text from images and videos in multiple languages, which is then indexed for search.
Additional system capabilities include:
- Support for both local and cloud-based AI engines.
- Pre- and post-processing functionality for AI data workflows.
- Use of combined metadata from related digital assets to enhance AI processing (e.g., for multi-modal vectorisation across text, image, and audio).
- Modular architecture allowing for multiple AI engines per task type (e.g., different translation engines per language pair).
- Integration with various organisational platforms, enabling transfer and centralised management of digital assets.
Profuz LAPIS is currently used by broadcasters such as Canal+ in France, the Council of Europe, and Bulgarian National Radio.
“The architecture in Profuz LAPIS is built with the vision to allow for further additional data processing, transformation and aggregation modules to be included. The ability to be able to add AI-enabled data processing, transformation and aggregation is already effortless in LAPIS, as it has been programmed to do it naturally. In LAPIS, this processing is classified and it’s possible to have more than one module per type,” said Profuz Digital CTO and developer Kamen Ferdinandov.
“For example, you can add multiple translation engines, so that the most appropriate one, based on language pair, or type of translation, and so forth, can be selected.”