Device for speech recognition with dictionary updating

• US 6,732,074 B1
• Filed: 01/27/2000
• Issued: 05/04/2004
• Est. Priority Date: 01/28/1999
• Status: Expired due to Term

• Current Assignee: Ricoh Company Limited
• Original Assignee: Ricoh Company Limited
• Inventors: Kuroda, Masaru
• Primary Examiner(s): Chawan, Vijay
• Assistant Examiner(s): Storm, Donald L.
• Application Number:
  US09/492,280
• Time in Patent Office: 1,559 Days
• Field of Search: 704/244, 704/249, 704/254, 704/10, 704/243
• US Class Current: 704/244
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 2015/0635 updating or merging of old and new templates; Mean values; Weighting

Abstract

A standard dictionary; a feature extracting unit which extracts features from an input speech; a matching unit which performs matching between the features of the input speech extracted by the feature extracting unit and the standard dictionary; a result outputting unit which outputs a matching result in the matching unit; and a dictionary updating portion which updates the standard dictionary are provided. The standard dictionary is built initially as a dictionary to be used for recognizing speeches produced by any independent speaker; and the dictionary updating unit updates the standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker based on the result of matching between the features extracted from the input speech and the standard dictionary.

Claims

1. A device for speech recognition comprising:

a standard dictionary;

a feature extracting unit which extracts features from an input speech;

a matching unit which performs matching of the features of the input speech extracted by said feature extracting unit against said standard dictionary;

a result outputting unit which outputs a matching result in said matching unit; and

a dictionary updating portion which updates said standard dictionary,

wherein;

said standard dictionary is built initially as a dictionary to be used for recognizing speeches produced by any independent speaker;

said dictionary updating portion updates said standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker based on the result of matching of the features extracted from the input speech against said standard dictionary; and

said standard dictionary is built initially as a dictionary, to be used for recognizing speeches produced by any independent speaker, as a result of standard features of each string of characters being disintegrated into phoneme units, the thus-obtained features of the respective phonemes being used as phoneme information, and the connection of the phonemes being used as path information;

said matching unit, when comparing features of input phonemes determined from the features extracted from the input speech for a string of characters with the phoneme information in said standard dictionary corresponding to said string of characters, performs evaluation of phoneme distance between the features of the input phonemes and the phoneme information in said standard dictionary corresponding to said string of characters; and

said dictionary updating portion, based on the result of said evaluation of phoneme distance, updates the phoneme information in said standard dictionary corresponding to said string of characters, and, thus, updates said standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker, wherein said dictionary updating portion updates the phoneme information in said standard dictionary corresponding to said string of characters, and, thus, updates said standard dictionary, only when the phoneme distance between the features of the input phonemes and the phoneme information in said standard dictionary corresponding to said string of characters exceeds a predetermined threshold as a result of said evaluation of phoneme distance, such that said updating portion does not update said standard dictionary when the phoneme distance between the features of the input phonemes and the phoneme information in said standard dictionary corresponding to said string of characters does not exceed said predetermined threshold.

2. A device for speech recognition comprising:

a standard dictionary;

a feature extracting unit which extracts features from an input speech;

a matching unit which performs matching of the features of the input speech extracted by said feature extracting unit against said standard dictionary;

a result outputting unit which outputs a matching result in said matching unit; and

a dictionary updating portion which updates said standard dictionary,

wherein;

said standard dictionary is built initially as a dictionary to be used for recognizing speeches produced by any independent speaker;

said dictionary updating portion updates said standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker based on the result of matching of the features extracted from the input speech against said standard dictionary; and

said standard dictionary is built initially as a dictionary, to be used for recognizing speeches produced by any independent speaker, as a result of standard features of each string of characters being disintegrated into phoneme units, the-thus-obtained features of the respective phonemes being used as phoneme information, and the connection of the phonemes being used as path information;

said matching unit, when comparing features of input phonemes determined from the features extracted from the input speech for a string of characters with the phoneme information in said standard dictionary corresponding to said string of characters, performs evaluation of phoneme distance between the features of the input phonemes and the phoneme information in said standard dictionary corresponding to said string of characters; and

said dictionary updating portion, based on the result of said evaluation of phoneme distance, updates the phoneme information in said standard dictionary corresponding to said string of characters, and, thus, updates said standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker, wherein said dictionary updating portion updates the phoneme information in said standard dictionary corresponding to the vowels of said string of characters, and, thus, updates said standard dictionary, only when the phoneme distance between the features of the input phonemes and the phoneme information in said standard dictionary corresponding to said string of characters exceeds a predetermined threshold as a result of said evaluation of phoneme distance, such that said dictionary updating portion does not update said standard dictionary when the phoneme distance between the features of the input phonemes and the phoneme information in said standard dictionary corresponding to said string of characters does not exceed said predetermined threshold.

3. A device for speech recognition comprising:

standard dictionary means;

feature extracting means for extracting features from an input speech;

matching means for performing matching of the features of the input speech extracted by said feature extracting means against said standard dictionary means;

result outputting means for outputting a matching result in said matching means; and

dictionary updating means for updating said standard dictionary means only when the difference between the features of the extracted input speech and said standard dictionary means exceeds a predetermined threshold,

wherein;

said standard dictionary means is built initially as dictionary means to be used for recognizing speeches produced by any independent speaker; and

when the difference between the features of the extracted input speech and said standard dictionary means exceeds said threshold, said dictionary updating means updates said standard dictionary means so as to provide dictionary means to be used for recognizing speeches produced by a dependent speaker based on the result of matching of the features extracted from the input speech against said standard dictionary means.

4. A method of conducting speech recognition, comprising the steps of:

a) extracting features from an input speech;

b) performing matching of the features of the input speech extracted in said step a) against a standard dictionary;

c) outputting a matching result of said step b); and

d) updating said standard dictionary only when the phoneme distance between the features of the extracted input speech and the standard dictionary exceeds a predetermined threshold,

wherein;

said standard dictionary is built initially as a dictionary to be used for recognizing speeches produced by any independent speaker; and

said step d) comprises the step of updating said standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker based on the result of matching of the features extracted from the input speech against said standard dictionary.

5. A machine-readable memory medium having a program embodied therein for causing a computer to perform a speech recognition, said program comprising:

a standard dictionary;

a feature extracting unit configured to extract features from an input speech;

a matching unit configured to perform matching of the features of the input speech extracted by said feature extracting unit against said standard dictionary;

a result outputting unit configured to output a matching result in said matching unit; and

a dictionary updating portion configured to update said standard dictionary only when the phoneme distance between the features of the extracted input speech and the standard dictionary exceeds a predetermined threshold,

wherein;

said standard dictionary is built initially as a dictionary to be used for recognizing speeches produced by any independent speaker; and

said dictionary updating portion updates said standard dictionary so as to provide a dictionary to be used for recognizing speeches produced by a dependent speaker based on the result of matching of the features extracted from the input speech against said standard dictionary.