Speech recognition method and apparatus for recognizing phonemes using a plurality of speech analyzing and recognizing methods for each kind of phoneme

Claim 3

A method for speech recognition according to claim 1, wherein said determining step determines the recognition method in accordance with each kind of phoneme in each segment.

Claim 4

A method for speech recognition according to claim 1, wherein said selecting step selects words in accordance with each segment of the words stored in the dictionary.

Claim 5

A method for speech recognition according to claim 1, further comprising the step of matching the input speech with the selected words, wherein said matching step matches each segment of the input speech with the segments of the selected words.

Claim 6

A method for speech recognition according to claim 1, wherein said recognizing step is performed by a central processing unit.

Claim 7

A method for speech recognition according to claim 1, wherein said selecting step selects words from the dictionary in which the phonemes of the words in the dictionary are stored in accordance with a segmentation operation which classifies each phoneme of the words stored in the dictionary wherein said method further comprises the steps of: matching the input speech with the selected words using normalized values; and outputting the result of said matching step.

Claim 8

A method for speech recognition according to claim 1, further comprising the step of transmitting the segments to a recognition device for recognizing phonemes by using the applied recognition methods.

Claim 11

An apparatus for speech recognition according to claim 8, wherein each segment comprises at least one phoneme, wherein said apparatus further comprises means for determining the kind of each phoneme, and wherein said determining means determines the recognition method in accordance with each kind of phoneme in each segment.

Claim 12

An apparatus for speech recognition according to claim 8, wherein said selecting means selects words in accordance with each segment of the words stored in the word dictionary.

Claim 13

An apparatus for speech recognition according to claim 8, further comprising means for matching of the input speech with the selected words, wherein said matching means matches each segment of the input speech with segments of the selected words.

Claim 14

An apparatus for speech recognition according to claim 8, wherein said recognition means comprises a central processing unit.

Claim 15

An apparatus for speech recognition according to claim 8, wherein said selecting means selects words from the dictionary in which the phoneme of words in the dictionary are stored in accordance with a segmentation operation which classifies each phoneme of the words in the dictionary, wherein said apparatus further comprises: matching means for matching the input speech with the selected words using converted values; and outputting means for outputting the result of matching by said matching means.

Claim 16

An apparatus for speech recognition according to claim 8, further comprising: transmitting means for transmitting the segments to said recognizing means for recognizing phonemes by using the applied recognition methods.

Claim 17

A computer program product comprising a computer usable medium having computer readable program code means embodied in the medium for speech recognition by a computer comprising a speech recognition apparatus, said product including: first computer readable program code means for causing a computer to input speech, which is to be recognized, into the speech recognition apparatus; second computer readable program code means for causing the computer to perform a segmentation operation which classifies the input speech into segments, each segment including phonemes; third computer readable program code means for causing the computer to determine the kinds of classes of phonemes in the input speech by referring to the segmentation operation; fourth computer readable program code means for causing the computer to apply only one recognition method to each class of phoneme caused to be determined by said third computer readable program means, wherein said fourth computer readable program code means causes the computer to apply different recognition methods to different classes of phonemes; fifth computer readable program code means for causing the computer to recognize phonemes in each segment by using the applied only one recognition method, and obtaining values of each phoneme as the recognition results; sixth computer readable program code means for causing the computer to convert the values for the segments into values of a predetermined range; seventh computer readable program code means for causing the computer to select words from a dictionary which are coincident with a sequence of the recognized phonemes: eighth computer readable program code means for causing the computer to calculate a score by summing up the values of phonemes of each selected word; and ninth computer readable program code means for causing the computer to output a word with a maximum score caused to be calculated by said eighth computer readable program code means as a recognition result of the input speech.

Claim 18

A computer program product comprising a computer usable medium having computer readable program code means embodied in the medium for speech recognition by a computer comprising a speech recognition apparatus, said product including: first computer readable program code means for causing a computer to input speech, which is to be recognized, into the speech recognition apparatus; second computer readable program code means for causing the computer to perform a segmentation operation which classifies the input speech into segments, each segment including phonemes; third computer readable program code means for causing the computer to determine the kinds of classes of phonemes in the input speech by referring to the segmentation operation; fourth computer readable program code means for causing the computer to apply only one recognition method to each class of phoneme caused to be determined by said third computer readable program means, wherein said fourth computer readable program code means causes the computer to apply different recognition methods to different classes of phonemes; fifth computer readable program code means for causing the computer to recognize phonemes in each segment by using the applied only one recognition method, and obtaining values of each phoneme as the recognition results; sixth computer readable program code means for causing the computer to convert the values for the segments into values of a predetermined range; seventh computer readable program code means for causing the computer to select words from a dictionary which are coincident with a sequence of the recognized phonemes: eighth computer readable program code means for causing the computer to calculate a score by summing up the values of phonemes of each selected word; and ninth computer readable program code means for causing the computer to output a word with a maximum score caused to be calculated by said eighth computer readable program code means as a recognition result of the input speech.