| 摘要 |
PROBLEM TO BE SOLVED: To provide a character input device which can be used immediately by a beginner, without depending on character inputting by ten keys of a portable telephone for dialing or using a keyboard of normal alphabets or the Japanese syllabaries. SOLUTION: Characters are inputted, based on a common operating section for all of the Japanese syllabaries and a single detection related to the syllabaries. Sound, having a contracted sound, is treated as one character, and a sound having a double consonants is treated as an independent single character. A common operating section is provided for all of the alphabets, besides the use of the Japanese syllabaries. Besides the Japanese syllabaries, an operating section is provided for inputting phrase dots or a reading dot and these dots are inputted without relating them to voice detection. The operations of the operating sections and the uttering of voice are stored, and character recognition is conducted, even if the operations of the sections and the recognition of a character deviate from each other. Voice patterns detected during a misrecognition are stored, and recognition accuracy is improved, by learning the correlation between the characters to be correctly recognized and the stored voice patterns. A phrase storage section is provided to store recognized character strings, and the recognition precision is improved by estimating the characters to be recognized next, from the character strings which are stored in the phrase storage section. |
