← Back to Patent Explorer
US5040213A Expired / Lapsed US ⏱️ 0.0 years remaining

Method of renewing reference pattern stored in dictionary

Assignee: Ricoh Co Ltd
•
Filing: 1990-01-22
•
Granted: 1991-08-13
•
Expires: 2010-01-22
📄 Original PDF 🌐 Google Patents 🔄 Sync Now

Abstract

Abstract A reference pattern renewing method includes inputting an input pattern of an utterance of a word, comparing the input pattern with reference patterns stored in a dictionary memory to thereby generate at least a first candidate having the highest degree of similarity and a second candidate having the second highest degree of similarity, and calculating a ratio of the highest degree of similarity to the second highest degree of similarity. The method further includes determining whether or not the ratio is equal to or less than a predetermined threshold value, calculating, from the input pattern, a renewed reference pattern to be substituted for a corresponding one of the reference patterns when it is determined that the ratio of the highest degree of similarity to the second highest degree of similarity is neither equal to nor less than the predetermined threshold value, and registering the renewed reference pattern in the dictionary memory instead of the corresponding one of the reference patterns.

Inventors

  • 👤 Seigou Yasuda
  • 👤 Kazuhiko Nishimura

Bibliographic Details

Application Number —
Jurisdiction US
Data Source google_patents
Last Synchronized 2026-11-05 17:19
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/06; G10L15/063

Claims Analysis

Total Claims: 15
Claim 1
Independent Claim 161 words
A reference pattern renewing method comprising: (a) inputting an input pattern of an utterance of a word; (b) comparing said input pattern with reference patterns stored in a dictionary memory to thereby generate at least a first candidate having the highest degree of similarity and a second candidate having the second highest degree of similarity; (c) calculating a ratio of the highest degree of similarity to the second highest degree of similarity; (d) determining whether or not said ratio is equal to or less than a predetermined threshold value; (e) calculating, from said input pattern, a renewed reference pattern to be substituted for a corresponding one of said reference patterns when it is determined that the ratio of the highest degree of similarity to the second highest degree of similarity is neither equal to nor less than said predetermined threshold value; and (f) registering said renewed reference pattern in said dictionary memory instead of said corresponding one of said reference patterns.
Claim 2
Dependent Claim 161 words
A reference pattern renewing method comprising: (a) inputting an input pattern of an utterance of a word; (b) comparing said input pattern with reference patterns stored in a dictionary memory to thereby generate at least a first candidate having the highest degree of similarity and a second candidate having the second highest degree of similarity; (c) calculating a ratio of the highest degree of similarity to the second highest degree of similarity; (d) determining whether or not said ratio is equal to or less than a predetermined threshold value; (e) calculating, from said input pattern, a renewed reference pattern to be substituted for a corresponding one of said reference patterns when it is determined that the ratio of the highest degree of similarity to the second highest degree of similarity is neither equal to nor less than said predetermined threshold value; and (f) registering said renewed reference pattern in said dictionary memory instead of said corresponding one of said reference patterns.
Claim 3
Dependent Claim 68 words
A reference pattern renewing method as claimed in claim 1, further comprising a step of determining whether the second degree of similarity is equal to or less than a second threshold value, wherein said step for calculating said renewed reference pattern calculates said renewed reference pattern when it is determined that the second highest degree of similarity is neither equal to nor less than the second threshold value.
Claim 4
Dependent Claim 170 words
A reference pattern renewing method as claimed in claim 1, further comprising the steps of: inputting a plurality of input patterns; determining whether a number of times that the corresponding one of said reference patterns is renewed in the past is equal to or less than a number of patterns used for creating said corresponding one of said reference patterns; creating a provisional reference pattern from said input patterns when it is determined that the number of times that the corresponding one of said reference patterns is renewed in the past is equal to the number of patterns used for the corresponding one of said reference patterns, said provisional reference pattern having a voiced section; detecting a silent section in said voiced section of said provisional reference pattern; and dividing said provisional reference pattern into blocks by said silent section to thereby generate a finalized reference pattern, wherein said step for registering said renewed reference pattern registers, as said renewed reference pattern, said finalized reference pattern in said dictionary memory.
Claim 5
Dependent Claim 106 words
A reference pattern renewing method as claimed in claim 1, further comprising the steps of: creating a provisional reference pattern from said input pattern; calculating a length R of a renewed reference pattern created from said provisional reference pattern by using the following formula, R=D(W-1)+X where W is a number of said patterns used for creating the corresponding one of said reference patterns, D is a length of the corresponding one of said reference patterns, and X is a length of said input pattern; and generating said renewed reference pattern having said length R to be registered in said dictionary memory from said provisional reference pattern.
Claim 6
Dependent Claim 119 words
A reference pattern renewing method as claimed in claim 1, further comprising the steps of: creating a provisional reference pattern from said input pattern; calculating a length R of a voiced section of a renewed reference pattern created from said provisional reference pattern by using the following formula, R=D(W-1)+X where W is the number of said patterns used for creating the corresponding one of said reference patterns, D is a length of a corresponding voiced section of the corresponding one of said reference pattern, and X is a length of a corresponding voiced section of said input pattern; and generating said renewed reference pattern having said voiced section having said length R to be registered in said dictionary memory.
Claim 7
Dependent Claim 152 words
A reference pattern renewing method as claimed in claim 1, further comprising the steps of: calculating an estimate value Xn indicating hour reliable of the corresponding one of said reference patterns in accordance with the following formula, Xn=X.sub.n-1 ·(1-m)+K·m where n is the number of times that a decision on whether the reference pattern should be renewed on the basis of said step (d) is made, X n-1 is a previous estimate value, m (0<m<1) is a renewal coefficient, and K is a value which assumes `0`when it is determined that the one of said reference patterns should be renewed and which assumes `0`when it is determined that one of said the reference patterns should not be renewed; and comparing said estimate value X with a predetermined reference value; and determining whether or not said renewed reference pattern should be registered in said dictionary memory on the basis of said estimate value.
Claim 8
Dependent Claim 23 words
A reference pattern renewing method as claimed in claim 6, wherein said predetermined threshold value is changed in accordance with said estimate value.
Claim 9
Dependent Claim 186 words
A reference pattern renewing method comprising: (a) inputting an input pattern of an utterance of a word; (b) comparing said input pattern with reference patterns stored in a dictionary memory to thereby generate at least a first candidate having the highest degree of similarity and a second candidate having the second highest degree of similarity; (c) calculating a ratio of the highest degree of similarity to the second highest degree of similarity; (d) determining whether or not said ratio is equal to or less than a predetermined threshold value; (e) making a predetermined additional decision on whether or not one of said reference patterns relating to the highest degree of similarity is to be renewed when said step (d) determines that said ratio is equal to or less than said predetermined threshold ratio; (f) calculating, from said input pattern, a renewed reference pattern to be substituted for said one of said reference patterns when said step (f) determines that said one of reference patterns should be renewed; and (g) registering said renewal reference pattern in said dictionary memory instead of said one of said reference patterns.
Claim 10
Dependent Claim 186 words
A reference pattern renewing method comprising: (a) inputting an input pattern of an utterance of a word; (b) comparing said input pattern with reference patterns stored in a dictionary memory to thereby generate at least a first candidate having the highest degree of similarity and a second candidate having the second highest degree of similarity; (c) calculating a ratio of the highest degree of similarity to the second highest degree of similarity; (d) determining whether or not said ratio is equal to or less than a predetermined threshold value; (e) making a predetermined additional decision on whether or not one of said reference patterns relating to the highest degree of similarity is to be renewed when said step (d) determines that said ratio is equal to or less than said predetermined threshold ratio; (f) calculating, from said input pattern, a renewed reference pattern to be substituted for said one of said reference patterns when said step (f) determines that said one of reference patterns should be renewed; and (g) registering said renewal reference pattern in said dictionary memory instead of said one of said reference patterns.
Claim 11
Dependent Claim 107 words
A reference pattern renewing method as claimed in claim 8, further comprising the steps of: calculating an average of a predetermined number of past degrees of similarity obtained in the past by said step (b); and determining whether a ratio of the highest degree of similarity to said average is equal to or less than a second predetermined threshold value, wherein when the ratio of the highest degree of similarity to said average is neither equal to nor less than said second predetermined threshold value, said step (f) calculates said renewed reference pattern to be substituted for said one of said reference patterns from said input pattern.
Claim 12
Dependent Claim 128 words
A reference pattern renewing method as claimed in claim 8, further comprising the step of: comparing said input pattern with said renewed reference pattern to thereby generate a degree of similarity; multiplying said degree of similarity to said renewed reference pattern by a predetermined constant to generate a multiplied degree of similarity to said renewed reference pattern; and determining whether said multiplied degree of similarity relating to said renewed reference pattern is equal to or less than a second predetermined threshold value; wherein when said multiplied degree of similarity to said renewed reference pattern is neither equal to nor less than said second predetermined threshold value, said step (f) calculates said renewed reference pattern to be substituted for said one of said reference patterns from said input pattern.
Claim 13
Dependent Claim 127 words
A reference pattern renewing method as claimed in claim 8, further comprising the steps of: calculating a number of voiced patterns included in said one of the reference patterns; calculating a number of voiced patterns included in said input pattern; calculating an absolute value of a difference between said number of voiced patterns in said one of the reference patterns and said number of voiced patterns included in said input pattern; and determining whether said absolute value is equal to or greater than a second predetermined threshold value, wherein when said absolute value is neither equal to nor greater than said second predetermined threshold value, said step (f) calculates said renewed reference pattern to be substituted for said one of said reference patterns from said input pattern.
Claim 14
Dependent Claim 111 words
A reference pattern renewing method as claimed in claim 8, comprising the steps of: calculating a length of the one of said reference patterns; calculating a length of said input pattern; calculating an absolute value of a difference between said length of the one of said reference patterns and said length of said input pattern; and determining whether said absolute value is equal to or greater than a second predetermined threshold value, wherein when said absolute value is neither equal to nor greater than said second predetermined threshold value, said step (f) calculates said renewed reference pattern to be substituted for said one of said reference patterns from said input pattern.
Claim 15
Dependent Claim 68 words
A reference pattern renewing method as claimed in claim 8, further comprising a step of determining whether the second degree of similarity is equal to or less than a second threshold value, wherein said step for calculating said renewed reference pattern calculates said renewed reference pattern when it is determined that the second highest degree of similarity is neither equal to nor less than the second threshold value.

References Cited (Backward Citations)

DE3216800A1 Cited
DE3129282A1 Cited
DE3129353A1 Cited
EP0191354A1 Cited
US4905287A Cited

USPTO Annuity & Maintenance Windows

Annuity Stage Window Opens Payment Deadline Surcharge Period Until Estimated Fee Status
3.5 Years (Year 4) 1994-08-12 1995-02-11 1995-08-13 $2000.00 Window Closed / Lapsed
7.5 Years (Year 8) 1998-08-12 1999-02-11 1999-08-13 $3760.00 Window Closed / Lapsed
11.5 Years (Year 12) 2002-08-12 2003-02-11 2003-08-13 $7700.00 Window Closed / Lapsed

Assigned Categories & Tags

No categories assigned to this patent yet.

Assign Category Tag