Voice recognition device, voice recognition method, and voice recognition program

• US 20080071536A1
• Filed: 09/11/2007
• Published: 03/20/2008
• Est. Priority Date: 09/15/2006
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Nagashima, Hisayuki
• Granted Patent: US 8,548,806 B2
• Time in Patent Office: Days
• 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 device, a voice recognition method and a voice recognition program capable of appropriately restricting recognition objects based on voice input from a user to recognize the input voice with accuracy are provided. The voice recognition device includes a first voice recognition processing unit for executing processing of assigning weights of a first ratio to a sound score and a language score calculated for the input voice and recognizing the voice using the obtained scores to determine a type of a domain representing the control object based on a result of the processing, and a second voice recognition processing unit, using the domain of the determined type as a recognition object, for executing processing of assigning weights of a second ratio to the sound score and the language score calculated for the input voice, the weight on the sound score being greater in the second ratio than in the first ratio, and recognizing the voice using the obtained scores to determine the control content of the control object based on a result of the processing.

Claims

1. A voice recognition device for determining a control content of a control object based on a recognition result of input voice, comprising:

a first voice recognition processing unit for executing processing on the input voice by assigning weights of a first ratio to a sound score calculated based on an acoustic feature of the voice and a language score calculated based on a linguistic feature of the voice, and recognizing the voice using the obtained scores, to determine a type of a domain representing the control object based on a result of the processing; and

a second voice recognition processing unit for executing processing, using the domain of the type determined by the first voice recognition processing unit as a recognition object, by assigning weights of a second ratio to the sound score and the language score calculated for the input voice, the weight on the sound score being greater in the second ratio than in the first ratio, and recognizing the voice using the obtained scores, to determine the control content of the control object based on a result of the processing.

2. The voice recognition device according to claim 1, wherein the first voice recognition processing unit and the second voice recognition processing unit execute the processing of recognizing the voice for the same voice.

3. The voice recognition device according to claim 1, comprising a control processing unit for executing prescribed control processing based at least on the control content of the control object determined.

4. The voice recognition device according to claim 3, comprising a response output processing unit for outputting a response to a user from which the voice is input, wherein

the control processing executed by the control processing unit includes processing of controlling the response to the user to prompt the user to input voice.

5. A voice recognition device including a microphone to which voice is input and a computer having an interface circuit for accessing data of the voice obtained via the microphone, and determining a control content of a control object through processing executed by the computer, based on a recognition result of the voice input to the microphone,

the computer executing:

sound score calculation processing of calculating a sound score based on an acoustic feature of the input voice;

language score calculation processing of calculating a language score based on a linguistic feature of the input voice;

for the input voice, first voice recognition processing of assigning weights of a first ratio to the sound score calculated in the sound score calculation processing and the language score calculated in the language score calculation processing, and recognizing the voice using the obtained scores;

domain type determination processing of determining a type of a domain representing the control object based on a recognition result in the first voice recognition processing;

for the input voice, second voice recognition processing, using the domain of the type determined in the domain type determination processing as a recognition object, of assigning weights of a second ratio to the sound score calculated in the sound score calculation processing and the language score calculated in the language score calculation processing, the weight on the sound score being greater in the second ratio than in the first ratio, and recognizing the voice using the obtained scores; and

recognition result determination processing of determining the control content of the control object based on a recognition result in the second voice recognition processing.

6. A voice recognition method for determining a control content of a control object based on a recognition result of input voice, comprising:

a first voice recognition step of assigning weights of a first ratio to a sound score calculated based on an acoustic feature of the input voice and a language score calculated based on a linguistic feature of the voice, and recognizing the voice using the obtained scores;

a domain type determination step of determining a type of a domain representing the control object based on a recognition result in the first voice recognition step;

a second voice recognition step, using the domain of the type determined in the domain type determination step as a recognition object, of assigning weights of a second ratio to the sound score and the language score calculated for the input voice, the weight on the sound score being greater in the second ratio than in the first ratio, and recognizing the voice using the obtained scores; and

a recognition result determination step of determining the control content of the control object based on a recognition result in the second voice recognition step.

7. A voice recognition program causing a computer to execute processing of determining a control content of a control object based on a recognition result of input voice, having a function to cause the computer to execute:

sound score calculation processing of calculating a sound score based on an acoustic feature of the input voice;

language score calculation processing of calculating a language score based on a linguistic feature of the input voice;

for the input voice, first voice recognition processing of assigning weights of a first ratio to the sound score calculated in the sound score calculation processing and the language score calculated in the language score calculation processing, and recognizing the voice using the obtained scores;

domain type determination processing of determining a type of a domain representing the control object based on a recognition result in the first voice recognition processing;

for the input voice, second voice recognition processing, using the domain of the type determined in the domain type determination processing as a recognition object, of assigning weights of a second ratio to the sound score calculated in the sound score calculation processing and the language score calculated in the language score calculation processing, the weight on the sound score being greater in the second ratio than in the first ratio, and recognizing the voice using the obtained scores; and

recognition result determination processing of determining the control content of the control object based on a recognition result in the second voice recognition processing.