Knowledge Engine (KE) System
The KE System is an NLU engine (and associated developer toolkit) for providing robust, low-cost, natural-speech call-routing for up to 15-20 routing destinations without the conventional need for transcriptions-based training data and high-priced SLM (statistical language model) grammar speech engines. It offers a powerful say-anything, open-prompt alternative to conventional discrete-speech, directed-dialog approach.
KE Mechanism
The KE System follows a three step process: (a) visual layout of the required menu-tree (ie. call-flow) using a graphical user interface; (b) specification of distinct keywords (or key-phrases) as synonyms for each node in the menu-tree; and (c) generation of natural-speech, say-anything recognizer and NLU grammars through combination of specified keywords with a unique domain and speaker-independent garbage model. The recognizer grammar is loaded inside the third-party recognizer and allows detection of specified keywords in a wide variety of utterances of speaker’s choice. The NLU grammar is loaded inside Vestec NLU engine and processes the output of the third-party recognizer for semantic interpretation. The KE system also provides full capability for testing and tuning natural-speech, say-anything recognizer and NLU grammars.
KE Benefits
The KE System represents a significant advance over conventional directed-dialog, discrete-speech call-routing solutions. The latter force callers to: (1) respond to system prompts in only specified keywords or short phrases, and (2) traverse a multi-level menu-tree one level at a time. By contrast, the KE system – for almost the same deployment time and cost – enables callers to: (3) respond to system prompts in natural speech utterances of their choice containing specified keywords, and (4) gain direct access to relevant node in a multi-level menu-tree with a single utterance. The KE System also allows replacement of directed-dialog prompts with an open-prompt (ie. general guidance, such as: “How can I help you?”).
