| 摘要 |
<p>PROBLEM TO BE SOLVED: To surely prepare the mail of contents desired by a speaking person even when being utilized by an unspecified speaking person. SOLUTION: Voice inputted from a portable telephone set 2 is voice- recognized by a mail preparing device 1 and it is converted into characters to prepare a mail sentence. In the device 1, a word list part 1i has a dialect dictionary and a vogue word dictionary in addition to the basic dictionary of each language. The dialect dictionary is prepared for each dialect grouped in each language, and a setting changing part 1e sets which dialect dictionary to use by instruction from the set 2. Concerning an acoustic model part 1g and a language model part 1h, ones different depending on a dialect are prepared, and the part 1e is set by the instruction from the set 2. In accordance with setting like this way, a sentence preparing part 1k prepares the mail sentence. The part 1i registers a corrected word corrected concerning the mail sentence by many persons in a dictionary and includes the vogue dictionary by which it is used preferentially in the mail sentence.</p> |
