Incorrect voice command recognition prevention and recovery processing method and apparatus

• US 5,577,164 A
• Filed: 01/23/1995
• Issued: 11/19/1996
• Est. Priority Date: 01/28/1994
• Status: Expired due to Term

• Current Assignee: Canon Ayutthaya Limited (Canon Inc.)
• Original Assignee: Canon Ayutthaya Limited (Canon Inc.)
• Inventors: Sakai, Keiichi, Kaneko, Kazue
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): Smits, Talivaldis Ivars
• Application Number:
  US08/376,913
• Time in Patent Office: 666 Days
• Field of Search: 395/2.6, 395/2.84, 395/825
• US Class Current: 704/275
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering

Abstract

Information for identifying whether a word should be recognized in the next recognition operation is stored in a recognition dictionary. This information is updated each time the recognized result is obtained in order to update the word for the next recognition, making it possible to prevent an incorrect recognition from being repeated. An indication of whether each word can be entered repeatedly is stored as identification information in the recognition dictionary. When the recognized word is a word which cannot be entered repeatedly, the identification information of the word is updated, thereby reducing the probability of incorrect recognition in the next recognition operation by excluding the word from the next recognition operation. Even when the recognized word is a word which can be entered repeatedly, the word is excluded from the next recognition operation when an instruction is issued from an instruction device. Thus, it is possible to reflect the will of the user in the recognition operation, and to decrease the probability of an incorrect recognition. The incorrect operation performed as a result of the incorrect recognition of the input information is recovered by an easy operation, and the burden on the user is lessened by preventing the repetition of an incorrect recognition operation.

Claims

1. An input information processing apparatus, comprising:

input means for inputting vocal information;

storing means for storing words to be recognized for voice recognition and for storing first identification information indicating whether each stored word should be made a candidate for a current voice recognition operation;

recognition means for selecting words to be determined to be a candidate for voice recognition on the basis of the first identification information by referring to said storing means and for recognizing vocal information input by said input means by selecting a candidate word determined to match the input vocal information as a recognition result;

output means for outputting the recognition result of the recognition operation performed by said recognition means;

indicating means for indicating and inputting into said apparatus an indication that the selected candidate word recognized as the recognition result outputted by said recognition means is an incorrect recognition; and

updating means for updating the first identification information on the selected candidate word indicated to be an incorrect recognition by changing the first identification information on the selected candidate word stored in said storing means to indicate that the selected candidate word will not be selected again in the recognition operation in accordance with an indication input from said indicating means.

2. An input information processing apparatus according to claim 1, wherein said storing means stores second identification information indicating whether each stored word can be entered more than once in succession in the voice recognition operation performed by said recognition means.

3. An input information processing apparatus according to claim 2, further comprising control means, when it is determined that a recognized word cannot be entered in the voice recognition operation more than once in succession by referring to the second identification information in said storing means, updating the second identification information so that second identification information indicates that the recognized word cannot be entered in the voice recognition operation regardless of the presence of absence of an indication from said indicating means.

4. An input information processing apparatus according to claim 1, wherein the vocal information is input over a public line into said input information processing apparatus.

5. An input information processing apparatus according to claim 1, further comprising display means for displaying characters representing the recognition result.

6. An input information processing apparatus, comprising:

input means for inputting vocal information;

storing means for storing words to be recognized for voice recognition and for storing first identification information indicating whether each stored word should be made a candidate for a current voice recognition operation;

recognition means for selecting words to be determined to be a candidate for voice recognition on the basis of the first identification information by referring to said storing means and for recognizing vocal information input by said input means by selecting a candidate word determined to match the input vocal information and a recognition result;

output means for outputting the recognition result of the recognition operation performed by said recognition means;

indicating means for indicating and inputting into said apparatus an indication that the selected candidate word recognized as the recognition, result outputted by said recognition means is an incorrect recognition; and

updating means for updating the first identification information on the selected candidate word of the recognition result stored in said storing means by changing the first identification information to information indicating that the selected candidate word of the recognition result will not be selected again in the recognition operation.

7. An input information processing apparatus according to claim 6, wherein said storing means stores second identification information indicating whether each stored word can be entered more than once in succession in the voice recognition operation, and wherein the second identification information is updated so that the second identification information indicates that the selected candidate word of the recognition result will not be selected again in the recognition operation when the second identification information of the selected candidate word of the recognition result indicates that the word cannot be entered more than once in succession in the recognition operation.

8. An input information processing apparatus according to claim 6, wherein the vocal information is input over a public line.

9. An input information processing apparatus according to claim 6, further comprising display means for displaying characters representing the recognition result.

10. An input information processing method, comprising the steps of:

inputting vocal information into an information processing apparatus;

recognizing voice information input in said inputting step by storing words for voice recognition and first identification information indicating whether each stored word should be a candidate for a current voice recognition operation, and by selecting a word from the stored words, determined to be a candidate for voice recognition on the basis of the first identification information to produce a recognition result;

outputting the recognition result of the recognition operation performed in said recognizing step;

indicating and inputting into the information processing apparatus an indication that the recognition result is an incorrect recognition; and

updating, in accordance with the input indication, the first identification information of the selected word indicated to be an incorrect recognition by changing the first identification information to information indicating that the selected word will not be selected again in said recognizing step.

11. An input information processing method according to claim 10, wherein said storing step stores second identification information indicating whether each stored word can be entered more than once ins succession in the voice recognition operation.

12. An input information processing method according to claim 11, further comprising the step of updating, when it is determined that a recognized word cannot be entered in the voice recognition operation more than once in succession by referring to the second identification information the second identification information so that the second identification information indicates that the recognized word cannot be entered in the voice recognition operation again regardless of the presence or absence of an indication from said indicating step.

13. An input information processing method according to claim 10, wherein said inputting step comprises the step of inputting the vocal information over a public line.

14. An input information processing method according to claim 10, further comprising the step of displaying characters representing the recognition result.

15. An input information processing method, comprising the steps of:

inputting vocal information into an information processing apparatus;

recognizing voice information input in said inputting step by storing words for voice recognition and first identification information indicating whether each stored word should be a candidate for a current voice recognition operation, and by selecting a word from the stored words determined to be a candidate for voice recognition on the basis of the first identification information to produce a recognition result;

outputting the recognition result of the recognition operation performed in said recognizing step;

indicating and inputting into the information processing apparatus an indication that the selected word of the recognition result is an incorrect recognition; and

updating the first identification information of the selected word indicated to be an incorrect recognition by changing the first identification information to information indicating that the selected word of the recognition result will not be selected again in the recognition operation.

16. An input information processing method according to claim 15, wherein said storing step stores second identification information indicating whether each stored word can be entered more than once in succession in the voice recognition operation, and wherein said method further comprises the step of updating the second identification information so that the second identification information indicates that the selected word of the recognition result will not be selected again in the recognition operation when the second identification information of the selected word of the recognition result indicates that the word cannot be entered more than once in succession in the recognition operation.

17. An input information processing method according to claim 15, wherein said inputting step comprises the step of inputting the vocal information over a public line.

18. An input information processing method according to claim 15, further comprising the step of displaying characters representing the recognition result.