A speech recognition method used in an information processing apparatus, the apparatus acquiring, from a search server, a search keyword list containing the search keywords searched by the search server and rank information associated with the search keyword list, selecting words from the search keyword list, acquiring, from a recognition word dictionary generating device that applies weight information for use in speech recognition to the selected words, the words having the weight information applied thereto, and performing a voice chat with other information processing apparatus, the method comprising the steps of:
- acquiring the words having the weight information applied thereto from the recognition word dictionary generating device and storing the words in a recognition database for use in speech recognition;
- obtaining voice data from the content of the conversation during the voice chat;
- analyzing the voice data based on an acoustic feature of the voice data and generating word sequence candidates corresponding to the content of the conversation;
- and
- analyzing the generated word sequence candidates based on a linguistic concatenation feature between the words that constitute the word sequence by referencing the recognition database, thereby selecting a most feasible word sequence,
- wherein the search server represents a website that provides a keyword search service and a directory search service so that a user uses the search server to search for information available through Internet.