Method and system for preventing entry of confusingly similar phases in a voice recognition system vocabulary list

• US 5,452,397 A
• Filed: 12/11/1992
• Issued: 09/19/1995
• Est. Priority Date: 12/11/1992
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: Texas Instruments, Inc.
• Inventors: Ittycheriah, Abraham P., Wheatley, Barbara J.
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): HAFIZ, TARIQ R
• Application Number:
  US07/989,285
• Time in Patent Office: 1,012 Days
• Field of Search: 381/41-42, 381/43, 381/46, 381/51, 395/2, 395/2.3, 395/2.4, 395/2.45, 395/2.48-2.49, 395/2.52-2.7, 395/2.84
• US Class Current: 704/240
• CPC Class Codes: G10L 15/07 to the speaker

Abstract

A method and system prevent the entry of confusingly similar phrases (60) in a vocabulary list (10) of a speaker-dependent voice recognition system. The method first receives (20, 30, 50) and enrolls the phrase (60) for adding to vocabulary list (10). Next, probabilities (62) are assigned to all phrases (12). Probabilities (62) assigned to already existing phrases are greater than the probability (64) assigned to the phrase (62) that the user wants to add. The phrase (60) to be added is received a second time and compared to the modified vocabulary list (61) that includes first-received phrase (60). Next, the method indicates whether the second-received phrase matches a phrase (12, 14, 16) on modified vocabulary list (61) other than first-received phrase (60).

Claims

1. A method for preventing the entry of confusingly similar phrases in a vocabulary list of a speaker-dependent voice recognition system, comprising the steps of:

first receiving a first-received phrase for adding to a plurality of other phrases on a vocabulary list;

enrolling said first-received phrase in said vocabulary list;

assigning a first probability to said phrase and a second probability to each of said plurality of other phrases, said second probability having a greater value than said first probability;

receiving a second-received phrase, said second-received phrase sounding essentially the same as said first-received phrase, said first probability and said second probability controlling the likelihood that said speaker-dependent voice recognition system will match said second-received phrase to said first-received phrase or said plurality of other phrases on said vocabulary list;

comparing said second-received phrase to said vocabulary list including said first-received phrase and said plurality of other phrases;

indicating whether said second-received phrase matches one of said plurality of other phrases; and

removing the enrollment of said first-received phrase to said vocabulary list in the event that said second-received phrase matches one of said plurality of other phrases.

2. The method of claim 1, wherein said comparing step is performed using a Hidden Markov Modeling process.

3. The method of claim 1, further comprising the step of maintaining said first-received phrase in said vocabulary list in the event that said comparing step matches said second-received phrase to said first-received phrase.

4. The method of claim 1, wherein said first probability maximizes the discrimination of similarities between said second-received phrase and each of said plurality of other phrases and minimizes the likelihood that said voice recognition system will match said second-received phrase with a different phrase from said first-received phrase.

5. The method of claim 1, wherein said indicating step comprises the step of playing a message stating that said first-received phrase is too similar to at least one of said plurality of other phrases.

6. The method of claim 1, wherein said second probability is a unitary probability.

7. The method of claim 1, wherein said enrolling step comprises the step of enrolling said first-received phrase using an energy-based end-pointing process.

8. A system for preventing the entry of confusingly similar phrases in a vocabulary list of a speaker-dependent voice recognition system, comprising:

receiving circuitry associated with said speaker-dependent voice recognition system for first receiving a first-received phrase for adding to a plurality of other phrases on a vocabulary list;

an enrolling mechanism for enrolling said phrase in said first-received vocabulary list;

assigning algorithm circuitry for assigning a first probability to said phrase and a second probability to each of said plurality of other phrases said second probability having a greater value than said first probability;

said receiving circuitry for further receiving a second-received phrase, said second-received phrase sounding essentially the same as said first-received phrase, said first probability and said second probability controlling the likelihood that said voice recognition system will match said second-received phrase to said first-received phrase or said plurality of other phrases on said vocabulary list;

comparison circuitry for comparing said second-received phrase to said vocabulary list, including said first-received phrase and said plurality of other phrases;

communication circuitry for indicating whether said second-received phrase matches one of said plurality of other phrases; and

circuitry for removing the enrollment of said first-received phrase to said vocabulary list in the event that said second-received phrase matches one of said plurality of other phrases.

9. The system of claim 8, wherein said comparison circuitry further comprises circuitry for performing said comparing using Hidden Markov Models.

10. The system of claim 8, further comprising maintaining circuitry for maintaining said first-received phrase in said vocabulary list in the event that said comparison circuitry matches said second-received phrase to said first-received phrase.

11. The system of claim 8, wherein said first probability maximizes the discrimination of similarities between said second-received phrase and each of said plurality of other phrases and minimizes the likelihood that said voice recognition system will match said second-received phrase with a different phrase from said first-received phrase.

12. The system of claim 8, wherein said communication circuitry further comprises playing circuitry playing a message stating that said first-received phrase is too similar to at least one of said plurality of other phrases.

13. The system of claim 8, wherein said second probability is a unitary probability.

14. The system of claim 8, wherein said enrolling mechanism comprises a means for enrolling said first-received phrase using energy-based end-pointing.

15. An improved speaker-dependent voice recognition system for preventing the entry of confusingly similar phrases in a vocabulary list, the system comprising:

a speaker-dependent voice recognition system for responding to vocal commands;

receiving circuitry associated with said speaker-dependent voice recognition system for first receiving a first-received phrase for adding to a plurality of other phrases on a vocabulary list;

an enrolling mechanism for enrolling said first-received phrase in said vocabulary list;

assigning algorithm circuitry for assigning a first probability to said phrase and a second probability to each of said plurality of other phrases said second probability having a greater value than said first probability;

said receiving circuitry for further receiving a second-received phrase, said second-received phrase sounding essentially the same as said first-received phrase, said first probability and said second probability controlling the likelihood that said voice recognition system will match said second-received phrase to said first-received phrase or said plurality of other phrases on said vocabulary list;

comparison circuitry for comparing said second-received phrase to said vocabulary list, including said first-received phrase and said plurality of other phrases; and

communication circuitry for indicating whether said second-received phrase matches one of said plurality of other phrases; and

rejection circuitry for removing the enrollment of said first-received phrase to said vocabulary list in the event that said second-received phrase matches one of said plurality of other phrases.

16. The improved system of claim 15, wherein said comparison circuitry further comprises circuitry for performing said comparing using a Hidden Markov Models.

17. The improved system of claim 15, further comprising maintaining circuitry for maintaining said first-received phrase in said vocabulary list in the event that said comparison circuitry matches said second-received phrase to said first-received phrase.

18. The improved system of claim 15, wherein said first probability maximizes the discrimination of similarities between said second-received phrase and each of said plurality of other phrases and minimizes the likelihood that said voice recognition system will match said second-received phrase with a different phrase from said first-received phrase.

19. The improved system of claim 15, wherein said communication circuitry further comprises playing circuitry playing a message stating that said first-received phrase is too similar to at least one of said plurality of other phrases.

20. The improved system of claim 15, wherein said second probability is a unitary probability.