VOICE RECOGNITION APPARATUS AND MEMORY PRODUCT

• US 20090150148A1
• Filed: 10/01/2008
• Published: 06/11/2009
• Est. Priority Date: 12/10/2007
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: ABE, Kenji
• Granted Patent: US 8,271,280 B2
• Time in Patent Office: Days
• US Class Current: 704/243
• CPC Class Codes: G10L 15/08 Speech classification or search

Abstract

A voice recognition apparatus can reduce false recognition caused by matching with respect to the phrases composed of a small number of syllables, when it performs a recognition process, by a pronunciation unit, for voice data based on voice produced by a speaker such as a syllable and further performs recognition by a method such as the Word Spotting for matching with respect to the phrases stored in the phrase database. The voice recognition apparatus performs a recognition process for comparing a result of the recognition process by a pronunciation unit with the extended phrases obtained by adding the additional phrase before and/or behind the respective phrases.

Claims

1. A voice recognition apparatus for performing a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, comparing a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and performing a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data,

said voice recognition apparatus comprising:

a priority recognition section for comparing the result of the first recognition process with a plurality of extended phrases, which are obtained by adding an additional phrase before and/or behind the respective phrases stored in the phrase database, to recognize the extended phrase included in the voice data in priority to the second recognition process.

2. The voice recognition apparatus according to claim 1, wherein

said priority recognition section gives priority to a recognition process related to the extended phrases by replacing the phrases stored in the phrase database by the extended phrases.

3. A voice recognition apparatus for performing a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, comparing a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and performing a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data,

said voice recognition apparatus comprising:

an extended phrase database for storing a plurality of extended phrases obtained by adding an additional phrase before and/or behind the respective phrases stored in the phrase database;

a section for extracting the phrases having a predetermined number of pronunciation units from the phrase database;

a section for extracting the extended phrases, which are obtained by adding an additional phrase to the extracted phrases, from the extended phrase database; and

a priority recognition section for comparing the result of the first recognition process with the extracted extended phrases to recognize the extended phrase included in the voice data in priority to the second recognition process.

4. The voice recognition apparatus according to claim 3, further comprising:

a setting section for setting said predetermined number based on the number of pronunciation units and the number of phrases, which are matched with any one of the phrases stored in the phrase database, in the result of the first recognition process.

5. The voice recognition apparatus according to claim 3, wherein

said extended phrase database stores a plurality of extended phrases obtained by adding a suffix and/or a prefix as the additional phrase to respective stems which are the respective phrases stored in the phrase database.

6. The voice recognition apparatus according to claim 3, wherein

said priority recognition section gives priority to a recognition process related to the extended phrases by replacing the phrases stored in the phrase database by the extended phrases.

7. The voice recognition apparatus according to claim 4, wherein

said setting section sets the number of pronunciation units, a total number of phrases matched with any one of the phrases stored in the phrase database counted for which exceeds a preset threshold, or a value smaller than or equal to said number of pronunciation units as the predetermined number in the first recognition process.

8. The voice recognition apparatus according to claim 4, wherein

said setting section sets the number of pronunciation units of the phrases, a total number of phrases matched with any one of the phrases stored in the phrase database counted for which exceeds a preset threshold, or a value smaller than or equal to said number of pronunciation units as the predetermined number in the first recognition process.

9. The voice recognition apparatus according to claim 4, wherein

said setting section counts a total number of phrases matched with any one of the phrases stored in the phrase database for each number of pronunciation units and sets the number of pronunciation units, the number of phrases counted for which is larger than or equal to a threshold preset based on a number counted for a phrase having other number of pronunciation units, or a value smaller than or equal to said number of pronunciation units as the predetermined number in the first recognition process.

10. A voice recognition apparatus for performing a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, comparing a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and performing a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data,

said voice recognition apparatus comprising:

a controller capable of performing operations of;

adding an additional phrase before and/or behind the respective phrases stored in the phrase database; and

comparing the result of the first recognition process with a plurality of extended phrases, which are obtained by adding the additional phrase, to recognize the extended phrase included in the voice data in priority to the second recognition process.

11. A voice recognition apparatus for performing a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, comparing a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and performing a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data,

said voice recognition apparatus comprising:

an extended phrase database for storing a plurality of extended phrases obtained by adding an additional phrase before and/or behind the respective phrases stored in the phrase database; and

a controller capable of performing operations of;

extracting the phrases having a predetermined number of pronunciation units from the phrase database;

extracting the extended phrases, which are obtained by adding an additional phrase to the extracted phrases, from the extended phrase database; and

comparing the result of the first recognition process with the extracted extended phrases to recognize the extended phrase included in the voice data in priority to the second recognition process.

12. A voice recognition apparatus for performing a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, comparing a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and performing a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data,

said voice recognition apparatus comprising:

priority recognition means for comparing the result of the first recognition process with a plurality of extended phrases, which are obtained by adding an additional phrase before and/or behind the respective phrases stored in the phrase database, to recognize the extended phrase included in the voice data in priority to the second recognition process.

13. A voice recognition apparatus for performing a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, comparing a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and performing a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data,

said voice recognition apparatus comprising:

an extended phrase database for storing a plurality of extended phrases obtained by adding an additional phrase before and/or behind the respective phrases stored in the phrase database;

means for extracting the phrases having a predetermined is number of pronunciation units from the phrase database;

means for extracting the extended phrases, which are obtained by adding an additional phrase to the extracted phrases, from the extended phrase database; and

priority recognition means for comparing the result of the first recognition process with the extracted extended phrases to recognize the extended phrase included in the voice data in priority to the second recognition process.

14. A computer-readable memory product storing a computer-executable computer program,

said computer program comprising:

a step of causing a computer to perform a first recognition process to recognize, by a pronunciation unit, voice data based on vocalization, compare a result of the first recognition process with a plurality of phrases stored in a phrase database, which stores a plurality of phrases having one or a plurality of pronunciation units, and perform a second recognition process to recognize a phrase, which is composed of one or a plurality of pronunciation units, included in the voice data; and

a step of causing the computer to compare the result of the first recognition process with a plurality of extended phrases, which are obtained by adding an additional phrase before and/or behind the respective phrases stored in the phrase database, to recognize the extended phrase included in the voice data in priority to the second recognition process.