System and method for modifying a language model and post-processor information

• US 7,774,196 B2
• Filed: 09/29/2004
• Issued: 08/10/2010
• Est. Priority Date: 10/01/2003
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Dictaphone Corporation (Microsoft Corporation)
• Inventors: Chadbourne, Jeffrey, Cote, William F.
• Primary Examiner(s): Albertalli; Brian L
• Application Number:
  US10/953,471
• Publication Number:
  US 20050165598A1
• Time in Patent Office: 2,141 Days
• Field of Search: None
• US Class Current: 704/9
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 2015/227 of the speaker; Human-factor methodology
• CPC Class Codes: G10L 2015/228 of application context

Abstract

A system and method for automatically modifying a language model and post-processor information is provided. The system includes the steps of language model identification, site-specific model creation, and language model adaptation.

Claims

1. A method in a computer system for selecting a language model, the method comprising:

operating the computer system to perform steps of;

collecting a plurality of language models;

collecting a set of corrected dictated reports;

comparing the corrected dictated reports with the plurality of language models;

determining a perplexity factor for each of the plurality of language models based on the comparison with the corrected dictated reports; and

selecting the one of the plurality of language models having the lowest perplexity factor.

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

collecting site specific data from a user of said computer system; and

creating a site-specific language model by adding the site specific data to the selected one of the plurality of language models having the lowest perplexity factor.

3. The method according to claim 2, wherein the site specific data comprises add word lists and post processing information.

4. The method according to claim 3, further comprising adapting the site-specific language model to create a task language model, and outputting the task language model.

5. The method according to claim 4, wherein adapting the site-specific language model comprises:

updating a set of statistics for the site-specific language model; and

adding an accent word list to the site-specific language model to create the task language model.

6. The method according to claim 5, wherein the set of statistics includes bigrams or trigrams.

7. The method according to claim 1, further comprising adapting the selected one of the plurality of language models with the lowest perplexity factor to create a task language model, and outputting the task language model.

8. The method according to claim 7, wherein adapting the selected one of the plurality of language models with the lowest perplexity factor comprises:

updating a set of statistics for the selected one of the plurality of language models; and

adding an accent word list to the selected one of the plurality of language models to create the task language model.

9. The method according to claim 8, wherein the set of statistics includes bigrams or trigrams.

10. A method in a computer system comprising:

operating the computer system to perform steps of;

selecting a language model from among a plurality of factory language models;

collecting site specific data from a user;

transforming the selected language model into a site-specific language model by adding the site-specific data to the selected language mode;

determining whether the user has an accent;

creating an accent specific word list based on the user'"'"'s accent; and

adding the accent specific word list to the site-specific language model.

11. The method according to claim 10, further comprising;

adapting the site-specific language model to create a task language model; and

outputting the task language model.

12. The method according to claim 11, wherein adapting the site-specific language model further comprises:

updating a set of statistics for the site-specific language model.

13. The method according to claim 12, wherein the set of statistics includes bigrams and/or trigrams.

14. A computer system comprising:

at least one computer programmed to perform steps of;

selecting a language model from among a plurality of factory language models;

collecting site specific data from a user;

transforming the selected language model into a site-specific language model by adding the site specific data to the selected language mode;

determining whether the user has an accent;

creating an accent specific word list based on the user'"'"'s accent; and

adding the accent specific word list to the site-specific language model.

15. The computer system according to claim 14, wherein the at least one computer is further programmed to:

adapt the site specific language model to create a task language model; and

output the task language model.

16. The computer system according to claim 15, wherein adapting the site-specific language model further comprises:

updating a set of statistics for the selected language model.

17. The computer system according to claim 16, wherein the set of statistics includes bigrams or trigrams.