Speech recognition system for languages with compound words

• US 5,754,972 A
• Filed: 05/10/1995
• Issued: 05/19/1998
• Est. Priority Date: 03/06/1992
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Dragon Systems, Inc. (Microsoft Corporation)
• Inventors: Bamberg, Paul G., Huang, Caroline B., Ellermann, Claudia L. E., VanEven, Stijn, Baker, James K., Roberts, Jed M.
• Primary Examiner(s): Tung, Kee M.
• Application Number:
  US08/437,965
• Time in Patent Office: 1,105 Days
• Field of Search: 395/2, 395/2.6, 395/2.64-2.66, 364/419.02, 364/419.06, 381/41, 381/43, 704/200, 704/251, 704/255-257, 704/2, 704/6
• US Class Current: 704/200
• CPC Class Codes: G10L 15/04 Segmentation; Word boundary detection
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 2015/085 Methods for reducing search complexity, pruning
• CPC Class Codes: G10L 2015/088 Word spotting

Abstract

A system and associated methods for recognizing compound words from an utterance containing a succession of one or more words from a predetermined vocabulary. At least one of the words in the utterance is a compound word including at least two formatives in succession, wherein those formatives are words in the vocabulary.

Claims

1. A system for recognizing compound words from an utterance representative of a succession of one or more formatives from a predetermined vocabulary, wherein each of said formatives has an associated acoustic model stored in memory, comprising:

A. a word recognizer including means;

i. selectively operative in a first mode for matching formatives of said utterance with said associated acoustic models according to a score representative of the likelihood of a formative being included in a compound word with other said formatives, for identifying said acoustic models and said associated matched formatives that are representative of a single compound word, and for combining said formatives in response to said identification for generating a single compound word,

ii. selectively operative in a second delimiting mode for matching said spoken formatives of said utterance with said associated acoustic models for identifying a Begin Compound Mode command which indicates the beginning of a compound word and an End Compound Mode command which indicates the end of a compound word, and for combining in response to the identification of said Begin Compound Mode command and said End Compound Mode command formatives spoken between said Begin Compound Mode command and said End Compound Mode command into a single compound word, and

B. a user interface, including means for selecting between said first mode and said second mode.

2. A system according to claim 1 wherein said user interface includes a microphone and is responsive to a user generated auditory signal to select between said first mode and said second mode.

3. A system according to claim 1 wherein said user interface includes a keyboard and is responsive to a user generated keyboard command to select between said first mode and said second mode.

4. A system according to claim 1 wherein said word recognizer in said second mode is selectively operative in a sub-mode for identifying as a character in said compound word keywords spoken between said first delimiter and said second delimiter, wherein each of said keywords is representative of a correspondingly positioned character in said compound word.

5. A system according to claim 4 wherein one or more of said keywords are International Communication Alphabet words.

6. A system according to claim 1 wherein said word recognizer is selectively operative in a third mode for identifying as a character in said compound word keywords entered by a user between said first delimiter and said second delimiter, wherein each of said keywords is representative of a correspondingly positioned character in said compound word.

7. A system according to claim 6 wherein said user interface includes a microphone and said keywords are entered through said microphone.

8. A system according to claim 6 wherein said user interface includes a keyboard and said key words are entered through said microphone.

9. A system according to claim 6 wherein one or more of said keywords are International Communication Alphabet words.

10. A system according to claim 1 wherein said word recognizer includes means for recognizing said compound words from said utterance, wherein said utterance contains one or more words spoken in isolation.

11. A system according to claim 1 wherein said word recognizer includes means for recognizing said compound words from said utterance, wherein said utterance contains one or more words spoken as continuous speech.

12. A system according to claim 1 wherein said means for identifying said acoustic models that are matched to successively spoken formatives that are representative of a single compound word performs said identification based at least in part on a language score which corresponds to a language model probability estimate of the probability that said single compound word would occur given a language context derived from one or more of said successively spoken formatives.

13. A system according to claim 1 wherein said means for identifying said acoustic models that are matched to successively spoken formatives that are representative of a single compound word performs said identification based at least in part on a language score which corresponds to a language model probability estimate of the probability that said single compound word would occur given a language context derived from one or more formatives which have preceded said single compound word.

14. A computer-implemented method for recognizing compound words from an utterance representative of a succession of one or more formatives from a predetermined vocabulary, wherein each of said formatives has an associated acoustic model stored in memory, comprising the steps of:

A. recognizing said compound words by;

i. selectively operating said computer in a first mode to match each of said spoken formatives to an associated acoustic model, to automatically identify ones of said acoustic models and said associated matched formatives that are representative of a single compound word according to a score representative of the likelihood of a formative being included in a compound word with other said formatives, and to combine said formatives in response to said identification,

ii. selectively operating said computer in a second delimiting mode to match one of said acoustic models to each of said spoken formatives to identify a spoken Begin Compound Mode command which indicates the beginning of a compound word and a spoken End Compound Mode command which indicates the end of a compound word, and to combine a succession of formatives matched to ones of said associated acoustic models spoken between said Begin Command Mode command and said End Compound Mode command into a single compound word.

15. The method according to claim 14 including the further step of selecting between said first and second modes in response to a user generated auditory signal.

16. The method according to claim 14 including the further step of selecting between said first and second modes in response to a user generated keyboard command.

17. The method according to claim 14 wherein operation in said second mode includes the further step of selectively operating in a sub-mode to identify as a character in said compound word keywords spoken between said first delimiter and said second delimiter, wherein each of said keywords is representative of a correspondingly positioned character in said compound word.

18. The method according to claim 17 wherein one or more of said keywords are International Communication Alphabet words.

19. The method according to claim 14 including the further step of selectively operating in a third mode to identify as a character in said compound word keywords entered by a user between said first delimiter and second delimiter, wherein each of said keywords is representative of a correspondingly positioned character in said compound word.

20. The method according to claim 19 including the further step of entering said keywords through user generated auditory signals.

21. The method according to claim 19 including the further step of entering said keywords through user generated keyboard commands.

22. The method according to claim 19 wherein one or more of said keywords are International Communication Alphabet words.

23. The method according to claim 14 including the further step of speaking one or more of said words as isolated words.

24. The method according to claim 14 including the further step of speaking one or more of said words as continuous speech.