LumenVox Announces Release of Speech Engine Version 10.1 Including Updated Voice Activity Detection Module

San Diego, CA (APRIL 4, 2011) — LumenVox, an innovator in speech recognition technology, has released LumenVox Speech Engine version 10.1, now supporting an updated Voice Activity Detection module, in addition to other new features and major improvements.

LumenVox's Speech Engine version 10.1 includes a new Voice Activity Detection (VAD 3.0) module that performs significantly better at distinguishing speech from non–speech, offering better overall performance. The VAD module 3.0 was implemented to greatly improve barge–in performance, allowing users to more accurately control the noise and speech levels when triggering barge–in.

"One of the key challenges for any speech recognition product is distinguishing between background noise and human speech," said LumenVox President Edward Miller. "The latest revision of our Voice Activity Detection (VAD) module in the 10.1 release of the LumenVox Speech Engine greatly improves the Engine's performance at detecting speech, allowing for faster and more accurate barge–in and end–of–speech detection, which in turn leads to more reliable and effective speech applications for our users."

Additional new features and major improvements to LumenVox's new Speech Engine Include:

• SSML editor view, allowing text–to–speech audio to be tuned within the Speech Tuner environment
• New acoustic model offering support for Indian English (en–IN)
• Virtual Machine (VM) environments that are now explicitly supported by the LumenVox License Server
• New Streaming options page in the Speech Tuner
• Added support for Multi–Part grammars in the Media Server
• Improved performance in critical low memory situations