Voice recognition system and method for recognizing input voice information

• US 7,809,562 B2
• Filed: 07/26/2006
• Issued: 10/05/2010
• Est. Priority Date: 07/27/2005
• Status: Expired due to Fees

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Hanazawa, Ken
• Primary Examiner(s): Hudspeth; David R
• Assistant Examiner(s): Rider; Justin W
• Application Number:
  US11/492,982
• Publication Number:
  US 20070027693A1
• Time in Patent Office: 1,532 Days
• Field of Search: 704/246
• US Class Current: 704/246
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems

Abstract

A voice recognition system has a recognition dictionary storing voice information, a primary voice recognition means for performing primary voice recognition in response to input voice information pronounced by a user by the use of the recognition dictionary, and a recognition result judging means for deciding whether the primary voice recognition result is to be accepted or rejected. The voice recognition system includes a transceiver means for sending the input voice information of the user to an additional voice recognition means when the primary voice recognition result is rejected by the recognition result decision means and for receiving a secondary voice recognition result produced as a result of secondary voice recognition of the additional voice recognition means, and a recognition result output means for outputting the primary or secondary voice recognition result to an exterior of the voice recognition system. The voice recognition system also includes a settled result input means for receiving settlement information on the primary or secondary voice recognition result and a dictionary content control means for updating said recognition dictionary based on the settlement information.

Claims

1. A voice recognition system for recognizing input voice information pronounced by a user and comprising:

a recognition dictionary for storing voice information;

a primary voice recognition unit which performs primary voice recognition of the input voice information by the use of said recognition dictionary to produce a primary voice recognition result which is specified by a voice feature of the input voice information;

a recognition result judgment unit which judges whether the primary voice recognition result is to be accepted or rejected;

a transceiver unit which sends the voice feature to additional voice recognition means for performing secondary voice recognition, when the primary voice recognition result specified by the voice feature is rejected by said recognition result judgment unit and for receiving a secondary voice recognition result produced by the additional voice recognition means;

a recognition result output unit which outputs the primary voice recognition result specified by the voice feature and outputted from said recognition result judgment unit or the secondary voice recognition result received by said transceiver unit to an exterior of said voice recognition system;

a settled result input unit which receives settlement information on the primary voice recognition result or the secondary voice recognition result outputted to the exterior of said voice recognition system; and

a dictionary content control unit which updates said recognition dictionary based on the settlement information inputted by said settled result input unit;

wherein;

said dictionary content control unit has a word history list which includes an order of use of each word and a frequency of use of each word and deletes, from the recognition dictionary, a word based on at least one of the oldest word and the smallest word of the frequency included in the word history list when an amount of the words in said recognition dictionary exceeds a processing capability of said voice recognition system.

2. The voice recognition system as claimed in claim 1, wherein the recognition result judgment unit comprises:

a recognition result presentation unit which presents the primary voice recognition result produced by said primary voice recognition means to an external decision system; and

a decision result input unit which receives a decision result to accept or reject the primary voice recognition result from the external decision system.

3. The voice recognition system recited in claim 1, wherein said primary voice recognition unit includes

(i) a voice pre-processing unit operable to analyze the input voice information so as to extract the voice feature thereof, and

(ii) a primary voice recognition unit operable to output the primary voice recognition result specified by the voice feature extracted by said voice preprocessing unit.

4. The voice recognition system recited in claim 1, wherein said dictionary content control unit is operable to send update information to the additional voice recognition means when said recognition dictionary is to be updated.

5. The voice recognition system as recited in claim 1, wherein the additional voice recognition unit is configured by a server device having a voice recognition function, the server device being communicable with said transceiver means via a communication network.

6. A voice recognition method comprising:

inputting input voice information pronounced by a user;

performing primary voice recognition of the voice information based on a recognition dictionary storing voice information to produce a primary voice recognition result specified by a voice feature of the input voice information;

deciding whether to accept or reject the primary voice recognition result specified by the voice feature;

sending the voice feature to additional voice recognition means for performing secondary voice recognition, when the primary voice recognition result specified by the voice feature is rejected;

receiving a secondary voice recognition result produced by the additional voice recognition means;

outputting the primary voice recognition result or the secondary voice recognition result to an exterior of a voice recognition system;

inputting settlement information in response to the primary voice recognition result or the secondary voice recognition result outputted to the exterior of the client terminal device

updating the recognition dictionary based on the inputted settlement information by the use of a word history list which includes an order of use of each word and a frequency of use of each word; and

deleting, from the recognition dictionary, a word based on at least one of the oldest work and the smallest word of the frequency included in the word history list when an amount of the words in said recognition dictionary exceeds a processing capability of said client terminal device.

7. The voice recognition method recited in claim 6, wherein the deciding step comprises:

presenting the primary voice recognition result to an external decision system; and

receiving a decision result to accept or reject the primary voice recognition result from the external decision system.

8. The voice recognition method claimed in claim 6, wherein the word history list is accessed through a communication network to send information on the word history list from the client terminal device to the exterior.

9. The voice recognition method recited in claim 6, wherein said updating the recognition dictionary comprises:

sending update information to the additional voice recognition means when updating said recognition dictionary.

10. A non-transitory computer-readable storage medium having a program recorded thereon for executing a procedure with a computer, said procedure comprising:

inputting a voice information of a user;

performing primary voice recognition of the voice information based on a recognition dictionary storing voice information to produce a primary voice recognition result of the input voice information;

judging whether to accept or reject the primary voice recognition result;

sending the input voice information of the user to additional voice recognition means for performing secondary voice recognition when the primary voice recognition result is rejected;

receiving a secondary voice recognition result produced by the additional voice recognition means;

outputting the primary voice recognition result or the secondary voice recognition result to an exterior of a voice recognition system;

inputting settlement information on the primary voice recognition

result or the secondary voice recognition result outputted to the exterior of the voice recognition system;

updating the recognition dictionary based on the inputted settlement information by the use of a word history list which includes an order of use of each word and a frequency of use of each word, and

deleting, from the recognition dictionary, a word based on at least one of the oldest word and the smallest word of the frequency included in the word history list when an amount of the words in said recognition dictionary exceeds a processing capability of said voice recognition system.

11. The non-transitory computer-readable storage medium claimed in claim 10, wherein the judging step comprises the steps of:

presenting the primary voice recognition result to an external decision system; and

receiving a decision result to accept or reject the primary voice recognition result from the external decision system.

12. The non-transitory computer-readable storage medium recited in claim 10, wherein said updating step of updating the recognition dictionary comprises sending update information to the additional voice recognition means when updating said recognition dictionary.