The system of searching audio messages for information by key words
In search settings user specifies the required phonograms of audio records and the list of words or word combinations.
The system supports two different search methods: index search – using full-text recognition results and acoustic search – by HMM (Hidden Marks Models) models which can be applied separately and combination with each other.
The system allows a user you to automate search by specified parameters for postprocessing of the results received. The solution can be deployed on a PC as an application or on a multiprocessor complex with client-server architecture.