The application provides a semantic engine-based intelligent navigation decoupling method for a call center, wherein after receiving real-time voice
stream of a user, a call center platform sends the real-time voice
stream to an ASR engine provided by an intelligent navigation basic capability provider through a standard media control protocol (MRCP protocol) to perform voice transcription and obtain text information; and the text information is routed to an external third-party semantic engine through a built-in and flexibly configurable semantic routing gateway for deep semantic understanding and dialogue decision-making, the third-party engine can be freely selected, a small model such as
natural language understanding (NLU) can be used, or a more advanced large
language model (LLM) can be used; after obtaining high-quality script text returned by the third-party engine, the call center calls a TTS service of the intelligent navigation basic capability provider to perform voice synthesis, so that the script is broadcast to the user through the call center media for listening. The application provides a new technical path for intelligent evolution of the call center.