Philips Launches SpeechPearl 8.0 Speech Recognition Engine

AACHEN, GERMANY and DALLAS, TX - Philips Electronics' (NYSE: PHG), a supplier of speech recognition technology and business solutions, announced the availability of its SpeechPearl 8.0 software package. SpeechPearl 8.0 is the successor of the Philips' telephony speech engine SpeechPearl 2000, containing new functionality to ease application development and integration, increase accuracy and improve application usability. SpeechPearl, a speech recognition and understanding software, enables IVR systems to understand people's natural speech through the spoken word. SpeechPearl, a speaker-independent speech recognition and natural language telephony engine, provides users with the latest technologies. Its 'open grammar' functionality allows natural language understanding for the caller and, at the same, application development for the system administrator. Developed for telephony speech applications, SpeechPearl provides speech recognition and understanding for human-like telephony services. It identifies spoken utterances that do not match the active grammar and rejects such utterances by reporting 'out-of-grammar' back to the application, thus avoiding wrong mapping of out-of-grammar utterances onto the active application grammar and increasing the flexibility and efficiency of closed grammar dialogues. More fluent natural language applications through intelligent grammar based barge-in SpeechPearl 8.0 allows for the design of more natural human-like dialogues through its robust grammar based barge-in feature. It separates human speech from background noise in the first pass via SpeechPearl's module, SpeechDetector™. In the second pass, the system only stops the dialogue if a caller's input reflects the context of the active application, thus controlling barge-in based on the user input against the active application grammar and improving the usability of natural language applications by providing a human-like dialogue flow. SpeechPearl 8.0 includes a new Voice Enrollment Framework feature that transparently and automatically handles the complete workflow of adding speaker-dependent recorded entries to a personal phone book. It not only registers names, but also provides clash detection against current entries in the personal phone book, as well as consistency checks of the sequence of recorded entries. It synchronizes a new entry with the existing database to avoid double entries or conflicts with similar names and checks name consistency. The new feature also enhances service quality for applications including personal directories, such as Voice Dialing and Auto Attendant. SpeechPearl 8.0 comes with a completely revised toolkit, which comprises a complete set of new tools and enhanced support of the entire application development process via a consistent and ergonomic workflow scheme. The new toolkit provides more graphical user interface (GUI) tools for application design, as well as for application tuning and evaluation. It incorporates tools for easily training with open grammars and for creating and transcribing speech corpuses. "For a number of years, Philips' speech recognition products have consistently evolved to handle the constantly changing application requirements," said Matthias Pankert, Director of Product Strategy and Planning at Philips Speech Processing/Telephony Solutions. "SpeechPearl 8.0 is the comprehensive and sophisticated upgrade to SpeechPearl 2000 with a variety of new features. Through SpeechPearl 8.0, we offer our customers the most advanced speech technology with outstanding ease of implementation and high adaptability. Integrating SpeechPearl 8.0 reduces maintenance and integration costs, since the migration to more sophisticated applications becomes simple and fast. Our new SpeechPearl release once again demonstrates Philips' leading position in cutting-edge, continuous speech recognition and natural language understanding technology."
SpeechTek Covers
Free
for qualified subscribers
Subscribe Now Current Issue Past Issues