Word-spotting speech recognition device and system

Claim 3

The device as claimed in claim 1 , further comprising a feature-extraction unit which extracts the features of input speech from the input speech without detecting speech frames, and wherein said matching unit compares the features of input speech with the features of the recognition objects so as to produce a similarity measure continuously without breaks of speech frames, and said dictionary updating unit updates the time lengths of phonemics when the similarity measure exceeds a predetermined threshold.

Claim 4

The device as claimed in claim 1 , wherein said dictionary updating unit compares a sum of the time lengths of phonemics constituting said one of the recognition objects with an actual time length of the input speech corresponding to said one of the recognition objects, and updates the time lengths of the phonemics in said dictionary based on a difference between the sum and the actual time length.

Claim 5

The device as claimed in claim 1 , wherein said dictionary updating unit obtains a difference between a time length of a phonemic in said dictionary and a time length of the same phonemic in the input speech, and adds the difference to the time length of the phonemic in said dictionary so as to update the time lengths of phonemics in said dictionary.

Claim 6

The device as claimed in claim 1 , wherein said dictionary updating unit obtains a difference between a time length of a phonemic in said dictionary and a time length of the same phonemic in the input speech, and adds the difference multiplied by a given weighting factor to the time length of the phonemic in said dictionary so as to update the time lengths of phonemics in said dictionary.

Claim 7

The device as claimed in claim 5 , wherein said weighting factor varies depending on how great the substantial similarities are.

Claim 8

The device as claimed in claim 5 , wherein said weighting factor varies depending on how many times the matching unit correctly finds the substantial similarities with respect to said one of the recognition objects.

Claim 9

The device as claimed in claim 1 , wherein said dictionary updating unit updates a time length of a given phonemic in said dictionary not only with respect to said one of the recognition objects but also with respect to all the recognition objects that include the given phonemic.

Claim 14

The method as claimed in claim 10 , further comprising a step of extracting the features of input speech from the input speech without detecting speech frames, and wherein said step b) compares the features of input speech with the features of the recognition objects so as to produce a similarity measure continuously without breaks of speech frames, and said step c) updates the time lengths of phonemics when the similarity measure exceeds a predetermined threshold.

Claim 15

The method as claimed in claim 10 , wherein said step c) compares a sum of the time lengths of phonemics constituting said one of the recognition objects with an actual time length of the input speech corresponding to said one of the recognition objects, and updates the time lengths of the phonemics in said dictionary based on a difference between the sum and the actual time length.

Claim 16

A machine-readable memory medium having a program embodied therein for causing a computer to perform a speech recognition, said program comprising: a codebook configured to store features of recognition objects and time lengths of phonemics, wherein the time lengths are an average time length defined for each phonemic; a matching unit configured to compare features of input speech with the features of the recognition objects; and a codebook updating unit configured to update the time lengths of phonemics in said codebook based on the input speech when said matching unit finds substantial similarities between the input speech and one of the recognition objects.

Claim 17

A machine-readable memory medium having a program embodied therein for causing a computer to perform a speech recognition, said program comprising: a codebook configured to store features of recognition objects and time lengths of phonemics, wherein the time lengths are an average time length defined for each phonemic; a matching unit configured to compare features of input speech with the features of the recognition objects; and a codebook updating unit configured to update the time lengths of phonemics in said codebook based on the input speech when said matching unit finds substantial similarities between the input speech and one of the recognition objects.

Claim 18

The machine-readable memory medium as claimed in claim 13 , wherein said program further comprises a feature-extraction unit configured to extract the features of input speech from the input speech without detecting speech frames, and wherein said matching unit compares the features of input speech with the features of the recognition objects so as to produce a similarity measure continuously without breaks of speech frames, and said dictionary updating unit updates the time lengths of phonemics when the similarity measure exceeds a predetermined threshold.

Claim 19

The machine-readable memory medium as claimed in claim 13 , wherein said dictionary updating unit compares a sum of the time lengths of phonemics constituting said one of the recognition objects with an actual time length of the input speech corresponding to said one of the recognition objects, and updates the time lengths of the phonemics in said dictionary based on a difference between the sum and the actual time length.