London, UK – April 7, 2008 – Fluency Voice Technology (http://www.fluencyvoice.com), the speech recognition solutions company, announced that they have been granted a patent on an innovative technology which is used in their solutions to improve the accuracy of speech recognition.

Speech recognizers are not perfect devices for converting spoken words into written words: they make numerous kinds of mistakes. Some of these mistakes are similar to the mistakes humans make when listening to other humans, such as confusing “thirty” for “thirteen”, while other types of mistakes are particular to the recognition operation and are quite unlike human-human errors, for example recognizing “January the 14th” as “March the 4th”. Researchers throughout the world are searching for improvements to the recognition algorithms to reduce the rate at which these mistakes occur, as it is these mistakes that cause the majority of the difficulty callers’ experience when using spoken-dialogue-driven systems.

This patent describes the improvements in speech recognition accuracy that can be obtained by introducing some distortions into the speech signal before it is sent to the recognizer. For this technique to be useable, the speech to be recognized is not just sent to the recognizer once, but multiple times, each time having been processed through a different distortion operation. After all of  the recognitions have been completed, the results for the various recognitions are compared to each other and an evaluation operation is performed to determine which of these results is most likely to be the correct one, and that is the one that is returned to the application. Fluency has applied this technique to all the leading speech recognizers currently on the market.

The result of this operation is that for a range of commonly occurring recognitions, the accuracy of the recognizer can be appreciably enhanced. Experiments have shown that the error in many common recognition tasks, such as recognizing postcodes, can be reduced by around 80%, and this reduction in error significantly enhances the feel of the dialogue to the caller and results in a higher call completion rate than would otherwise be achievable.

Dr Trevor Thomas, the inventor and Chief Scientist at Fluency, stated “This invention will deliver important improvements to recognition accuracy and will increase the performance of our spoken dialogue systems when compared to similar dialogue systems that just make conventional use of a speech recognizer”.

“Fluency continues to invest in R&D to improve the accuracy of speech recognition and provide superior applications that effectively manage the complexities involved in deploying the technology in order to provide optimal conversational caller experience.  This patent, one in a series of patents that Fluency has filed, demonstrates our commitment to excellence and our continued drive to innovation in this area” said Dr Jacob Abboud, Fluency’s CTO.

About Fluency Voice Technology

Fluency provides on-premise and hosted packaged speech recognition solutions for call centers. Fluency’s speech recognition solutions are used by call centers to improve customer service and significantly reduce costs. These call centers come to us because of the unique way we package and provide our speech solutions, which ensures a rapid, inexpensive, trouble-free rollout. Our configurable speech recognition solutions are delivered via the VSA™ Suite. VSA provides a range of applications which can be deployed individually or collectively to automate call center workflows. Each VSA deployment is fully configurable, allowing customers to make changes in response to business needs.

Fluency has offices in the UK and the USA. For more information please visit the corporate website on www.fluencyvoice.com.