Semantic Interfaces (SI) is a Natural Language Understanding (NLU) company. SI’s founders and employees are experts in various aspects of computational linguistics, computational semantics, general linguistics and Automatic Speech Recognition (ASR) technologies. SI is offering an NLU platform designated to comprehend ASR distorted speech commands and queries.

SI developed a unique NLU system designed to cope with a growing market pain: leading corporates across the world have recently launched a new trend named “Voice Activated Applications” – activating computer applications by natural language voice commands. The general impression out in the field is that voice activation doesn’t really work and most early adopter users drop their voice activated trials in no time. Indeed, none of the existing voice enabled products is considered successful.

User voice command is processed by one or more ASR engines, each returning a recognized text. SI’s NLU engine reconstructs the user’s actual command or query using its linguistic and semantic model of the domain. The textual command is then converted to a target application language.

The Company

  • Operating since 2010 in Tel-Aviv
  • Company established (& seed financing) at 2011
  • First project (2011): SiTEL – SMS medical simulation joint venture (supported by the BIRD Foundation)
  • Second project (2012): NDS – Cisco, voice activated TV
  • Developed SI’s NLU platform designated to translate free language commands into application-specific command language.
  • Developed a semantic Post ASR* enhancement technology
* Also referred to as VR or AVR (Automatic Voice Recognition)

The Forthcoming ASR Revolution

Speech recognition technology is becoming mature:

  • Voice activation is one of the most popular emerging trends these days
  • Market pain: current speech recognition technology doesn’t reach a satisfactory level of usability and is far from being commercial

Learn About Our Intuition

?   Question:

How can a mother understand her two year old baby when it asks for something even though more than 50% of the words it utters are illegible?

!   Answer:

By taking into account contextual consideration and by using linguistic intuitions.

What we do:

SI simulates human contextual and linguistic intuitions and reconstructs incoherent text commands into application specific coherent commands.

General block diagram

Leave a Reply