System for speech recognition

• US 20040243406A1
• Filed: 01/29/2004
• Published: 12/02/2004
• Est. Priority Date: 01/29/2003
• Status: Active Grant

• Current Assignee: Cerence Operating Company (Cerence Inc.)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Rinscheid, Ansgar
• Granted Patent: US 7,460,995 B2
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G10L 15/08 Speech classification or search

Abstract

This invention provides a system for speech recognition comparing speech against stored character strings in memory. Speech is transformed into spoken character strings. To accelerate the identification, a small group of characters from the stored character strings and the spoken character string are compared and the probabilities for identification may be calculated from those results. Those stored patterns, where the probability for identifying the speech exceeds a predetermined value, may be selected for further processing. The selected strings may have the remaining characters added to the group of characters for the next comparison. Alternatively, the number of characters for comparison may be incremented by a predetermined number in a step-by-step fashion, reducing the number of comparisons in subsequent steps as the probabilities for identification rise. The character string in memory with the highest probability for the identification match may be recognized as the character string identifying the speech.

Claims

1. A method for recognizing speech comprising:

comparing a predetermined number of spoken characters to a corresponding set of character strings stored in a memory;

selecting a group of character strings from the memory based on the first comparison;

comparing a second predetermined number of spoken characters to the corresponding characters in the selected group of character strings; and

identifying the spoken characters based upon the second comparison.

2. The method according to claim 1, where the corresponding character string is stored in the memory according to a tree structure.

3. The method according to claim 1, where the predetermined number of spoken characters is a portion of the total characters in a spoken character string.

4. The method according to claim 1, where the act of selecting further comprises calculating a matching probability based on the first comparison.

5. The method according to claim 1, where the second comparison further comprises comparing all of the spoken characters to the characters in the selected character strings.

6. The method according to claim 1, where the second comparison compares more characters than the first comparison.

7. The method according to claim 1, where the second comparison compares additional characters in incremental steps.

8. The method according to claim 1, where identifying further comprises recognizing a selected character string that has the highest probability of identifying the spoken character string.

9. A system for speech recognition comprising:

a microphone;

a processor for digitizing speech signals received from the microphone;

a memory coupled to the processor for storing character strings;

a comparator for comparing a spoken character string formed from the speech signals to the stored character strings in the memory; and

a controller programmed to select a group of stored character strings based on a comparison of a predetermined number of characters of the spoken character string to a predetermined number of stored characters and further programmed to identify a stored character string within the group of stored character strings that recognizes the spoken character string.

10. The system according to claim 9, where the stored character strings are stored in the memory according to a tree structure.

11. The system according to claim 9, where the predetermined number of spoken characters is a portion of the characters in the spoken character string.

12. The system according to claim 9, where the controller is further programmed to identify the spoken character string by comparing all of the spoken characters to the characters in the selected character strings.

13. The system according to claim 12, where all of the spoken characters are compared in incremental steps.

14. The system according to claim 9, where the controller is further programmed to select by calculating a matching probability based on the comparison.

15. The system according to claim 9, where the controller is further programmed to identify a stored character string within the group of stored character strings that has the highest probability of identifying the spoken character string.

16. A system for sound recognition comprising:

a microphone;

a processor for digitizing sound signals received from the microphone;

a memory coupled to the processor for storing character strings;

a comparator for comparing a sound character string formed from the sound signals to the stored character strings in the memory; and

a controller programmed to select a group of stored character strings based on a comparison of a predetermined number of characters of the sound character string to a predetermined number of stored characters and further programmed to identify a stored character string within the group of stored character strings that recognizes the sound character string.

17. A system according to claim 16, where the stored character strings are stored in the memory according to a tree structure.

18. The system according to claim 16, where the spoken characters are compared in incremental steps.

19. The system according to claim 16, where the controller is further programmed to select by calculating a matching probability based on the comparison.

20. The system according to claim 16, where the controller is further programmed to identify a stored character string within the group of stored character strings that has the highest probability of identifying the sound character string.