Speech recognition system

• US 7,392,182 B2
• Filed: 12/18/2002
• Issued: 06/24/2008
• Est. Priority Date: 12/18/2002
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Harman International Industries Incorporated (Samsung Electronics Co. Ltd.)
• Inventors: Gerl, Franz S., Enderle, Harald Bernd, Nuessle, Gerhard
• Primary Examiner(s): {hacek over (S)}mits; Talivaldis Ivars
• Assistant Examiner(s): SPOONER, LAMONT M
• Application Number:
  US10/324,737
• Publication Number:
  US 20040122670A1
• Time in Patent Office: 2,015 Days
• Field of Search: 704/231, 704/251
• US Class Current: 704/231
• CPC Class Codes: G10L 15/04 Segmentation; Word boundary detection

Abstract

A speech recognition system processes speech outputs from a user to select a list element from a list of list elements. The user speaks the speech output as a whole word and an initial character sequence of the desired list element. The whole word and individual character sequence components are processed separately. A sublist is generated in response to the individual character sequence. A list element is selected in response to the whole word and sublist.

Claims

1. A method for recognizing speech to select a list element from a list of list elements, comprising:

stating a speech output in chronologically successive voice inputs including the whole word of a desired list element as well as at least one initial character of the desired list element, wherein both the whole word and the at least one initial character are stated for every desired list element;

separating the speech output into whole word and individual character sequence components;

processing separately the whole word and individual character sequence components; and

performing speech recognition configured for the whole word in response to the individual character sequence.

2. The method according to claim 1, further comprising:

storing the whole word of the desired list element that was separated as a component of the speech output;

generating a sublist of the list elements from the list of list elements based on the individual character sequence component filtered out of the speech output;

comparing the list elements of the sublist with the stored whole word of the desired list element in the speech recognition of the two processed components of the speech output; and

generating a recognition result in response to the comparison of the list elements, where the recognition result is the list element most likely to match the desired list element.

3. The method according to claim 1, further comprising separating the speech output into the whole word and individual character components by means of speech recognition of individual characters.

4. The method according to claim 3, further comprising storing the portion of the speech output not associated with the individual character sequence as a whole word.

5. The method according to claim 1, where the two voice inputs of the speech output take place in chronological succession.

6. The method according to claim 5, where the two voice inputs of the speech output take place in chronological succession separated by a pause.

7. The method according to claim 1, the first voice input is the whole word of the desired list element, and the second voice input is at least one initial character of the desired list element.

8. The method according to claim 1, where the first voice input to be spoken is at least one initial character of the desired list element, and the second voice input to be spoken is the whole word of the desired list element.

9. The method according to claim 1, further comprising transmitting the list element most likely to match the desired list element as the recognition result.

10. The method according to claim 9, further comprising transmitting the list element acoustically.

11. The method according to claim 9, further comprising transmitting the list element optically.

12. The method according to claim 1, further comprising:

matching a hypothetical list with the list of list elements, the hypothetical list having a character sequence comprising at least one possible initial character of the desired list element; and

generating a sublist.

13. The method according to claim 12, further comprising:

filtering the desired list element out of the speech output; and

forming the character sequences of the hypothetical list from the component individual character sequences of the desired list element and from characters that are similar to the filtered-out individual characters.

14. The method according to claim 12, further comprising appending the character sequences of the hypothetical list to the list elements contained in the whole list.

15. The method according to claim 1 implemented in one of a destination guidance system and a navigation system.

16. A speech recognition system, comprising:

an input unit for voice input of a speech output comprising a whole word of a desired list element and at least one initial character of the desired list element, wherein both the whole word and the at least one initial character are stated for every desired list element;

a speech recognition unit for separating the speech output into whole word and individual character components and processing separately the whole word and individual character sequence components;

a memory unit for storing the whole word;

a database comprising a list of list elements for creating a sublist based on the match with the individual character;

a speech recognition unit configured with vocabulary created based on the list elements of the sublist for comparing the stored whole word of the desired list element with the list elements of the sublist; and

an output unit for transmitting the list element evaluated as being the most likely desired list element.

17. A method for recognizing speech to select a list element from a list of list elements, comprising:

separating a speech output into a whole word and individual character sequence components, wherein both the whole word and at least one initial character are stated for every speech output;

processing separately the whole word and the individual character sequence components;

storing the whole word;

generating a sublist in response to the individual character sequence; and

outputting a list element responsive to the whole word and the sublist.

18. The method according to claim 17, where the whole word and individual character sequence are chronologically successive voice inputs.

19. The method according to claim 18, where the whole word and individual character sequence are chronologically successive voice inputs separated by a pause.

20. The method according to claim 18, where the initial character sequence is the first voice input chronologically.

21. The method according to claim 18, where the whole word is the first voice input chronologically.

22. The method according to claim 17, further comprising where the individual character sequence comprises one initial character of the list element.

23. The method according to claim 17, further comprising where the individual character sequence comprises four initial characters of the list element.

24. The method according to claim 17, further comprising transmitting the list element acoustically.

25. The method according to claim 17, further comprising transmitting the list element optically.

26. The method according to claim 17, further comprising matching a hypothetical list with the sublist, the hypothetical list having a character sequence comprising at least the initial character of the list element.