Techniques to normalize names efficiently for name-based speech recognition grammars

• US 8,990,080 B2
• Filed: 01/27/2012
• Issued: 03/24/2015
• Est. Priority Date: 01/27/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Varkey, Mini, Sana, Bernardo, Boctor, Victor, Carlomagno, Diego
• Primary Examiner(s): Abebe, Daniel D
• Application Number:
  US13/359,919
• Publication Number:
  US 20130197906A1
• Time in Patent Office: 1,152 Days
• Field of Search: None
• US Class Current: 704/234
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules

Abstract

Techniques to normalize names for name-based speech recognition grammars are described. Some embodiments are particularly directed to techniques to normalize names for name-based speech recognition grammars more efficiently by caching, and on a per-culture basis. A technique may comprise receiving a name for normalization, during name processing for a name-based speech grammar generating process. A normalization cache may be examined to determine if the name is already in the cache in a normalized form. When the name is not already in the cache, the name may be normalized and added to the cache. When the name is in the cache, the normalization result may be retrieved and passed to the next processing step. Other embodiments are described and claimed.

Claims

1. A computer-implemented method, comprising:

receiving a name for normalization during a name-based speech grammar generation;

checking a normalization cache for the received name;

normalizing the received name and adding the normalization result to the normalization cache, when the received name is not in the normalization cache;

retrieving the normalization result from the normalization cache;

providing the normalization result to a next process in the name-based speech grammar generation;

maintaining a plurality of normalization caches, where each normalization cache corresponds to a different culture; and

providing a normalization result in a culture to the next process in a name-based speech recognition grammar generation specific to that culture.

2. The method of claim 1, further comprising:

dividing a received name into component names;

checking the normalization cache for each component name separately;

normalizing each name component separately and adding the normalization result component to the normalization cache, when the name component is not in the normalization cache;

retrieving the normalization result component from the normalization cache; and

providing the normalization result component to a next process in the name-based speech grammar generation.

3. The method of claim 1, further comprising:

persisting the normalization cache in a volatile memory.

4. The method of claim 1, further comprising:

persisting the normalization cache on a non-volatile computer-readable storage medium.

5. The method of claim 4, further comprising:

assigning an expiration date to the normalization cache; and

rebuilding the normalization cache when a date exceeds the expiration date.

6. The method of claim 4, further comprising:

comparing the names in a current version of a source of names to a version of the source of names from a time when the normalization cache was generated; and

rebuilding the cache when a difference between the versions of the source of names exceeds a threshold.

7. The method of claim 1, wherein a name comprises at least one of:

a component of a name, a full name, a person'"'"'s name, a number, an e-mail address, an internal address, a telephone number, a street address, a city name, a state name, a country name, a nickname, a suffix, a title, or an initial.

8. The method of claim 1 wherein each different culture is a different language culture.

9. An article comprising a computer readable storage device comprising instructions that when executed cause a system to:

check a normalization cache for a name received during a name-based speech grammar generation;

normalize the received name and add the normalization result to the normalization cache, when the received name is not in the normalization cache;

retrieve the normalization result from the normalization cache;

provide the normalization result to a next process in the name-based speech grammar generation;

maintain a plurality of normalization caches, where each normalization cache corresponds to a different culture; and

provide a normalization result in a culture to the next process in a name-based speech recognition grammar generation specific to that culture.

10. The article of claim 9, the medium further comprising instructions that when executed cause the system to:

divide a received name into component names;

checking the normalization cache for each component name separately;

normalize each name component separately and add the normalization result component to the normalization cache, when the name component is not in the normalization cache;

retrieve the normalization result component from the normalization cache; and

provide the normalization result component to a next process in the name-based speech grammar generation.

11. The article of claim 9, the medium further comprising instructions that when executed cause the system to:

persist the normalization cache on a non-volatile computer-readable storage medium.

12. The article of claim 11, the medium further comprising instructions that when executed cause the system to:

assign an expiration date to the normalization cache; and

rebuild the normalization cache when a date exceeds the expiration date.

13. The article of claim 11, the medium further comprising instructions that when executed cause the system to:

compare the names in a current version of a source of names to a version of the source of names from a time when the normalization cache was generated; and

rebuild the cache when a difference between the versions of the source of names exceeds a threshold.

14. The article of claim 9, wherein a name comprises at least one of:

a component of a name, a full name, a person'"'"'s name, a number, an e-mail address, an internal address, a telephone number, a street address, a city name, a state name, a country name, a nickname, a suffix, a title, or an initial.

15. The article of claim 9 wherein each different culture is a different language culture.

16. An apparatus, comprising:

a processing unit;

a memory to store a normalization cache;

a name normalizer executing on the processing unit to;

receive a name for normalization during a name-based speech grammar generation;

check a normalization cache for the received name;

normalize the received name and add the normalization result to the normalization cache, when the received name is not in the normalization cache;

retrieve the normalization result from the normalization cache;

provide the normalization result to a next process in the name-based speech grammar generation;

a plurality of normalization caches, where each normalization cache corresponds to a different culture; and

the name normalizer further to provide a normalization result in a culture to the next process in a name-based speech recognition grammar generation specific to that culture.

17. The apparatus of claim 16, wherein the memory is a volatile memory and the normalization cache is generated on-the-fly for each generation of a name-based speech grammar.

18. The apparatus of claim 16, wherein the memory is a non-volatile memory and the normalization cache is persisted for a plurality of generations of a name-based speech grammar.

19. The apparatus of claim 18, wherein the normalization cache is persisted until at least one of:

an expiration of the normalization cache, according to an assigned expiration date;

or

a difference between a current version of a source of names and a version of the source of names from a time when the normalization cache was generated exceeds a threshold.

20. The apparatus of claim 16 wherein each different culture is a different language culture.