Brainhat's dynamic semantic modeling system promises more accurate speech recognition.

Download the Brainhat Speech White Paper

©2003 Brainhat, Inc.
All rights reserved.

One of the most exciting areas for Brainhat is improved speech recognition. Because Brainhat is knowledge-based, it can anticipate what it might hear next, and use context to better improve recognition rates. For those recognition systems that require a context-free grammar, such as VoiceXML or telephony speech engines, Brainhat can generate dynamic grammars. At any time, one can ask Brainhat to list all the things it is expecting to hear.

  • Brainhat anticipates what the user might say before they say it.
  • Brainhat can use conceptual knowledge to understand what the user means.
  • Brainhat remembers previous conversations, as well as a current conversation which helps to grasp the topic of discussion.
  • Brainhat has the ability to switch topics mid-conversation.

How would one use Brainhat's ability to anticipate input, generate dynamic grammar and track a conversation to improve speech recognition? One straightforward use would be as a "shim" between speech engine and application:

Brainhat, Inc. sees improvements in speech recognition through the application of knowledge as one of the key components for bringing the recognition business forward, and for enabling free-form dialog between humans and machines.

Take a look at our technology. It could change the speech recognition world.