Apparatus and method for distinguishing similar-sounding utterances speech recognition

• US 6,487,532 B1
• Filed: 09/24/1998
• Issued: 11/26/2002
• Est. Priority Date: 09/24/1997
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: ScanSoft, Inc. n/k/a Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Gallopyn, Guido, Schoofs, Koen
• Primary Examiner(s): Knepper, David D.
• Application Number:
  US09/159,838
• Time in Patent Office: 1,524 Days
• Field of Search: 704/9, 704/235, 704/255-257, 704/270, 704/275, 704/270.1, 704/251, 704/252-254, 704/246-249
• US Class Current: 704/251
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering
• CPC Class Codes: G10L 2015/226 using non-speech characteristics

Abstract

Speaker-specified hints are used to establish conditions for a speech recognition system to select a recognition result for a previously provided utterance from among various possible homophones. The hints may characterize the utterance by a linguistic property, such as an orthographic, morphological, or semantic property.

Claims

1. A method of utilizing a speech recognizer to distinguish a provided utterance from one or more similar-sounding utterances when a speaker-specified hint as to the spelling of the provided utterance is provided, the hint being different from a full or partial spelling starting from the beginning of the provided utterance, the method comprising:

(a) identifying the hint and associating with it the provided utterance;

(b) using the hint to establish a condition for distinguishing the provided utterance from the one or more similar sounding utterances; and

(c) selecting a recognition result, derived in conjunction with operation of the speech recognizer, that satisfies the condition.

2. A method according to claim 1, wherein the step of selecting a recognition result, derived in conjunction with the operation of the speech recognizer, that satisfies the condition includes:

(a) providing a list of alternative recognition possibilities and

(b) filtering the list based upon the condition.

3. A method according to claim 2, wherein the step of providing a list of alternative recognition possibilities includes using the speech recognizer to provide entries in the list.

4. A method according to claim 2, wherein the step of providing a list of alternative recognition possibilities includes utilizing a linked dictionary to provide entries in the list.

5. A method according to claim 2, wherein the step of providing a list of alternative recognition possibilities includes both using the speech recognizer and utilizing a linked dictionary to provide entries in the list.

6. A method according to claim 1, wherein the hint includes reference to the context of previous dictation to characterize the provided utterance.

7. A method according to claim 1, wherein the hint is a linguistic property characterizing the provided utterance.

8. A method according to claim 7, wherein the hint is an orthographic property characterizing the provided utterance.

9. A method according to claim 7, wherein the hint is a morphological property characterizing the provided utterance.

10. A method according to claim 7, wherein the hint is a semantic property characterizing the provided utterance.

11. A method according to claim 1, wherein the hint is a fractional spelling of the provided utterance.

12. A method according to claim 1, wherein the hint is a grammatical term characterizing the provided utterance.

13. A method of utilizing a speech recognizer to distinguish a provided utterance from one or more similar-sounding utterances when a plurality of speaker-specified hints as to the spelling of the provided utterance is provided, the hint being different from a full or partial spelling starting from the beginning of the provided utterance, the method comprising:

(a) identifying the hints and associating with them the provided utterance;

(b) using the hints to establish conditions for distinguishing the provided utterance from the one or more similar-sounding utterances; and

(c) selecting a recognition result, derived in conjunction with the operation of the speech recognizer, that satisfies the conditions.

14. A speech recognition system for distinguishing a provided utterance from one or more similar-sounding utterances when a speaker-identified hint as to the spelling of the provided utterance is provided, the hint being different from a full or partial spelling starting from the beginning of the provided utterance, the system comprising:

(a) a hint recognizer which identifies the hint and associates with it the provided utterance;

(b) a condition specifier, coupled to the hint recognizer, which uses the hint to establish a condition for distinguishing the provided utterance from the one or more similar-sounding utterances; and

(c) a result selector, coupled to the condition specifier, which selects a recognition result that satisfies the condition.

15. A system according to claim 14, wherein the system further comprises a dictionary, to which the result selector is coupled, to provide entries in the list of alternative recognition possibilities.

16. A system according to claim 15, wherein the improvement further comprises a dictionary, to which the result selector is coupled, to provide entries in the list of alternative recognition possibilities.

17. A system according to claim 14, wherein the hint includes reference to the context of previous dictation to characterize the provided utterance.

18. A system according to claim 14, wherein the hint is a linguistic property characterizing the provided utterance.

19. A system according to claim 18, wherein the hint is an orthographic property characterizing the provided utterance.

20. A system according to claim 18, wherein the hint is a morphological property characterizing the provided utterance.

21. A system according to claim 18, wherein the hint is a semantic property characterizing the provided utterance.

22. A system according to claim 14, wherein the hint is a fractional spelling of the provided utterance.

23. A system according to claim 14, wherein the hint is a grammatical term characterizing the provided utterance.

24. A speech recognition system for distinguishing a provided utterance from one or more similar-sounding utterances when a plurality of speaker-specified hints as to the spelling of the provided utterance is provided, the hint being different from a full or partial spelling starting from the beginning of the provided utterance, the system comprising:

(a) a hint recognizer for identifying the hints and associating with them the provided utterance;

(b) a condition specifier, coupled to the hint recognizer, for using the hints to establish a condition for distinguishing the provided utterance from the one or more similar-sounding utterances; and

(c) a result selector, coupled to the condition specifier, for selecting a recognition result that satisfies the conditions.

25. A speech recognition system that is capable of distinguishing a provided utterance from one or more similar-sounding utterances when a speaker-specified hint as to the spelling of the provided utterance is provided, the hint being different from a full or partial spelling starting from the beginning of the provided utterance, the system comprising:

(a) means for identifying the hint and associating with it the provided utterance;

(b) means for using the hint to establish a condition for distinguishing the provided utterance from the one or more similar-sounding utterances; and

(c) means for selecting a recognition result, derived in conjunction with operation of the speech recognizer, that satisfies the condition.