Available Technology

Context-Driven Speech Recognition Software

The Context-Driven Speech Recognition Software is a patent-pending software application designed to handle very large complex vocabularies in smaller subsets by logically dividing the vocabulary according appropriate contexts. The divided vocabulary reduces the potential paths in the language model for any given recognition event, thereby reducing the recognition errors produced in the recognition engine. The framework for this software minimizes the size of the vocabulary loaded at any point when it is executing a recognition task against an audio stream. This basic principle alleviates many of the shortcomings related to off-the-shelf speech recognition engines’ handling of very large vocabularies. The approach manages speech recognition grammar outside of the compiled software. This application can be used for customer interface systems with large vocabularies, voice command & control applications, mobile applications (such as banking applications), computer gaming, and others.

Patent Abstract: 

The invention is system and method to recognize speech vocalizations using context-specific grammars and vocabularies. The system and method allow increased accuracy of recognized utterances by eliminating all language encodings irrelevant to the current context and allowing identification of appropriate context transitions. The system and method creates a context dependent speech recognition system with multiple supported contexts, each with specific grammar and vocabulary, and each identifying the potential context transition allowed. The system and method also include programmatic integration between the context dependent speech recognition system and other systems to make use of the recognized speech.

Inventors: 

Smith, Peter

Mouri, Ouns

Ouakil, Lisa

Ouakil, Abdelhamid

Patent Number: 
9460721
Patent Status: 
Patent Issued
Patent Issue Date: 
October 4, 2016
Share to Facebook Share to Twitter Share to Google Plus Share to Linkedin