Method and apparatus to use semantic inference with speech recognition systems

• US 20050038650A1
• Filed: 09/21/2004
• Published: 02/17/2005
• Est. Priority Date: 09/29/2000
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Bellegarda, Jerome R., Silverman, Kim E.A.
• Granted Patent: US 7,289,950 B2
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks

Abstract

A method and apparatus to use semantic inference with speech recognition systems includes recognizing at least one spoken word, processing the spoken word using a context-free grammar, deriving an output from the context-free grammar, and translating the output to a predetermined command.

Claims

1. A machine-implemented method for speech recognition comprising:

recognizing at least one spoken word;

processing said spoken word using a context-free grammar;

deriving an output from said context-free grammar; and

translating said output into a predetermined command.

2. The method of claim 1 further comprising performing an action corresponding to said predetermined command.

3. The method of claim 1 wherein said translating comprises semantically inferring said predetermined command from said output.

4. The method of claim 3 wherein said predetermined command is a semantic anchor.

5. The method of claim 1 wherein said output is a string of recognized words.

6. The method of claim 1 further comprising mapping said output to a vector space.

7. The method of claim 6 wherein said vector space is derived from training data.

8. The method of claim 6 further comprising determining at least one distance within said vector space between said output and a plurality of predetermined commands.

9. The method of claim 8 further comprising selecting the predetermined command that is closest in distance within said vector space to said output.

10. The method of claim 1 wherein said context-free grammar is an extended context-free grammar.

11. The method of claim 1 wherein:

said output comprises a plurality of words; and

selecting said predetermined command is not affected by the order of said words within said output.

12. A machine-readable medium having machine-executable instructions to a cause a machine to perform a method comprising:

recognizing at least one spoken word;

processing said spoken word using a context-free grammar;

deriving an output from said context-free grammar; and

translating said output into a predetermined command.

13. The machine-readable medium of claim 12 wherein the machine-executable instructions cause the machine to perform a method further comprising performing an action corresponding to said predetermined command.

14. The machine-readable medium of claim 12 wherein the machine-executable instructions cause the machine to perform a method wherein said translating comprises semantically inferring said predetermined command from said output.

15. The machine-readable medium of claim 14 wherein the machine-executable instructions cause the machine to perform a method wherein said predetermined command is a semantic anchor.

16. The machine-readable medium of claim 12 wherein the machine-executable instructions cause the machine to perform a method wherein said output is a string of recognized words.

17. The machine-readable medium of claim 12 wherein the machine-executable instructions cause the machine to perform a method further comprising mapping said output to a vector space.

18. The machine-readable medium of claim 17 wherein the machine-executable instructions cause the machine to perform a method wherein said vector space is derived from training data.

19. The machine-readable medium of claim 17 wherein the machine-executable instructions cause the machine to perform a method further comprising determining at least one distance within said vector space between said output and a plurality of predetermined commands.

20. The machine-readable medium of claim 19 wherein the machine-executable instructions cause the machine to perform a method further comprising selecting the predetermined command that is closest in distance within said vector space to said output.

21. The machine-readable medium of claim 12 wherein the machine-executable instructions cause the machine to perform a method wherein said context-free grammar is an extended context-free grammar.

22. The machine-readable medium of claim 12 wherein the machine-executable instructions cause the machine to perform a method wherein:

said output comprises a plurality of words; and

selecting said predetermined command is not affected by the order of said words within said output.

23. A machine-readable medium having machine-executable instructions to cause a machine to perform a method comprising:

recognizing at least one spoken word;

processing said spoken word using a context-free grammar;

deriving an output from said context-free grammar; and

converting said output to at least one predetermined command using semantic inference.

24. A machine comprising:

a processing unit;

a memory unit coupled to said processing unit by a system bus;

at least one machine-readable medium coupled to said processing unit by said system bus;

a speech recognition unit stored in said machine-readable medium;

a context-free grammar stored in said machine-readable medium; and

a semantic inference engine stored in said machine-readable medium.

25. The machine of claim 24 wherein said speech recognition unit comprises instructions causing said processing unit to recognize at least one spoken word.

26. The machine of claim 25 wherein said context-free grammar comprises instructions causing said processing unit to process said spoken word.

27. The machine of claim 26 wherein said context-free grammar comprises instructions causing said processing unit to derive an output.

28. The machine of claim 27wherein said output is a string of recognized words.

29. The machine of claim 27 wherein said semantic inference engine comprises instructions causing said processing unit to translate said output into a predetermined command.

30. The machine of claim 29 wherein said processor performs an action corresponding to said predetermined command.

31. The machine of claim 29 wherein said semantic inference engine further comprises instructions causing said processing unit to map said output to a vector space.

32. The machine of claim 31 wherein said vector space is derived from training data.

33. The machine of claim 31 wherein said semantic inference engine further comprises instructions causing said processing unit to determine at least one distance within said vector space between said output and a plurality of predetermined commands.

34. The machine of claim 31 wherein said semantic inference engine further comprises instructions causing said processing unit to select the predetermined command that is closest in distance within said vector space to said output.

35. The machine of claim 24 wherein said context-free grammar is an extended context-free grammar.

36. The machine of claim 27 wherein:

said output comprises a plurality of words; and

said predetermined command is not affected by the order of said words within said output.

37. A computerized system comprising:

means for recognizing at least one spoken word;

means for processing said spoken word;

means for deriving an output corresponding to said spoken word; and

means for converting said output to at least one predetermined command.

38. The machine of claim 37 further comprising means for executing an action corresponding to a predetermined command.