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

1. A voice recognition device comprising a microprocessor for determining a control object and control content thereof from predefined types of control objects and control contents thereof based on a recognition result obtained from input to a voice input unit, comprising:

a voice recognition unit for converting, via the microprocessor, the input verbal command into a text expressed with a series of words by recognizing the input verbal command;

a first parsing unit for performing an identification process of identifying a first control candidate group as a control candidate for the control object and control content thereof by parsing the text converted by the voice recognition unit;

a second parsing unit for performing an identification process of identifying a second control candidate group as a control candidate for the control object and control content thereof by parsing the text converted by the voice recognition unit with respect to control objects which are among the predefined types of control objects but not included in the first control candidate group; and

a control candidate identification unit for identifying a final control candidate group for determining the control object and control content thereof from the first control candidate group and the second control candidate group.

2. The voice recognition device as claimed in claim 1 , wherein the second parsing unit includes a calculation unit for calculating a word score based on features of the words contained in the text converted by the voice recognition unit, and identifies the second control candidate group by parsing the text with respect to the control objects which are among the predefined types of control objects but not included in the first control candidate group arid whose each word score satisfies a predefined condition.

3. The voice recognition device as claimed in claim 2 , wherein the second parsing unit calculates an average value of word scores for the words contained in the text and correspond to the control objects respectively, and uses the average value as the predefined condition if the average value is not smaller than a predefined threshold value.

4. The voice recognition device as claimed in claim 1 , further comprising a discrepancy judging unit for judging whether the control object and control content of each control candidate in the second control candidate group is contradictory to the control object and control content of the first control candidate group; and the second parsing unit excludes a control candidate from the second control candidate group if the control object and control content of the excluded control candidate is judged by the discrepancy judging unit to be contradictory to the control object and control content of the first control candidate group.

5. The voice recognition device as claimed in claim 1 , wherein the second parsing unit excludes a control candidate from the second control candidate group if the excluded control candidate can not be identified with a control object after the identification process performed on the second control candidate group.

6. A computer assisted voice recognition method for determining a control object and control content thereof from predefined types of control objects and control contents based on a recognition result obtained from a verbal command input to a voice input unit, the computer comprising a microprocessor, the method comprising:

a voice recognition step of converting, via the microprocessor, the input verbal command into a text expressed with a series of words by recognizing the input verbal command; a first parsing step of performing an identification process of identifying a first control candidate group as a control candidate for the control object and control content thereof by parsing the text converted in the voice recognition step;

a second parsing step of performing an identification process of identifying a second control candidate group as a control candidate for the control object and control content thereof by parsing the text converted in the voice recognition step with respect to control objects which are among the predefined types of control objects but not included in the first control candidate group; and

a control candidate identifying step of identifying a final control candidate group for determining the control object and control content thereof from the first control candidate group and the second control candidate group.

7. The voice recognition method as claimed in claim 6 , wherein the second parsing step includes a calculation step of calculating a word score based on features of the words contained in the text converted by the voice recognition step, and identifies the second control candidate group by parsing the text with respect to the control objects which are among the predefined types of control objects but not included in the first control candidate group and whose each word score satisfies a predefined condition.

8. The voice recognition method as claimed in claim 7 , wherein the second parsing step calculates an average value of word scores for the words contained in the text and correspond to the control objects respectively, and uses the average value as the predefined condition if the average value is not smaller than a predefined threshold value.

9. The voice recognition method as claimed in claim 6 , comprising a discrepancy judging step of judging whether the control object and control content of each control candidate in the second control candidate group is contradictory to the control object an control content of the first control candidate group; and the second parsing step excludes a control candidate from the second control candidate group if the control object and control content of the excluded control candidate is judged by the discrepancy judging step to be contradictory to the control object and control content of the first control candidate group.

10. The voice recognition method as claimed in claim 6 , wherein the second parsing step excludes a control candidate from the second control candidate group if the excluded control candidate cannot be identified with a control object after the identification process performed on the second control candidate group.

11. A non-transitory computer-readable storage medium with an executable voice recognition program stored thereon, wherein the program causes the computer to execute a process of determining a control object and control content thereof from predefined types of control objects and control contents based on a recognition result obtained from a verbal command input to a voice input unit, having a function to cause the computer to execute:

a voice recognition process for converting the input verbal command into a text expressed with a series of words by recognizing the input verbal command;

a first parsing process of performing an identification process of identifying a first control candidate group as a control candidate for the control object and control content thereof by parsing the text converted by the voice recognition process;

a second parsing process of performing an identification process of identifying a second control candidate group as a control candidate for the control object and control content thereof by parsing the text converted by the voice recognition process with respect to control objects which are among the predefined types of control objects but not included in the first control candidate group; and

a control candidate identification process of identifying a final control candidate group for determining the control object and control content thereof from the first control candidate group and the second control candidate group.