摘要 |
PURPOSE: A multi-service providing method in a voice recognition system is provided to integrate a variety of services into one system and to provide various information items to users. CONSTITUTION: A process is generated for each service and carried out in a separate process address space, and threads(201,202) are generated by channels in one process to share an HMM parameter. In one process, the threads share one HMM parameter but do not share word dictionary, grammar information and word list. The threads share the HMM parameter, word dictionary, grammar information and word list in one process. Multiple HMM parameters are used in one process, the threads are generated based on the HMM parameters, and the threads share the word dictionary, grammar information and word list. |