System and method for post processing speech recognition output

• US 20050108010A1
• Filed: 09/29/2004
• Published: 05/19/2005
• Est. Priority Date: 10/01/2003
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Dictaphone Corporation (Microsoft Corporation)
• Inventors: Santisteban, Ana, Frankel, Alan
• Granted Patent: US 7,996,223 B2
• Time in Patent Office: Days
• US Class Current: 704/235
• CPC Class Codes: G06F 40/103 Formatting, i.e. changing of presentation of documents automatic justification G06F40/189; automatic line break hyphenation G06F40/191
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

A system and method may be disclosed for facilitating the conversion of dictation into usable and formatted documents by providing a method of post processing speech recognition output. In particular, the post processing system may be configured to implement rewrite rules and process raw speech recognition output or other raw data according to those rewrite rules. The application of the rewrite rules may format and/or normalize the raw speech recognition output into formatted or finalized documents and reports. The system may thereby reduce or eliminate the need for post processing by transcriptionists or dictation authors.

Claims

1. A method for converting output from a speech recognition engine, the method comprising the steps of:

collecting an output from the speech recognition engine;

converting the output to a list of tokens;

identifying a first set of tokens matching a first set of predetermined patterns;

performing a first set of rewrite rules, where the first set of rewrite rules is performed to transform text that matches a first set of predetermined patterns;

identifying and transforming number clusters adjacent to keywords;

identifying and transforming number clusters not adjacent to keywords;

performing a second set of rewrite rules;

performing a third set of rewrite rules,

performing consolidation or suppression of filled pauses and silences

performing attachment and formatting of punctuation

performing capitalization of words following punctuation

2. The method according to claim 1, wherein the second set of rewrite rules may operate on the result of the first set of rewrite rules as well as the transformed number clusters.

3. The method according to claim 2, further comprising the step of performing at least a second rule interpretation performed to transform text that matches a second set of predetermined patterns.

4. The method according to claim 3, further comprising the step of converting punctuation tokens into symbols that cling to an adjacent word.

5. The method according to claim 4, where the adjacent word is capitalized.

6. The method according to claim 5, where filled pauses and silences are consolidated or suppressed and the boundaries between them and adjacent words are adjusted.

7. The method according to claim 6, further comprising the step of removing text used according to a set of post processor rules.

8. The method according to claim 7, further comprising the step of writing a file containing post-processed output to a target location.

9. The method according to claim 8, where the target location is a buffer.

10. The method according to claim 9, where the target location is a file.

11. A method for converting output from a speech recognition engine, the method comprising the steps of:

collecting output from a speech recognition engine whereby the output includes text having at least one of a token and a number cluster;

converting automatic speech recognition engine output to a list of tokens;

performing at least a first rule interpretation, where the rule interpretation is performed to transform text that matches a first set of predetermined patterns;

wherein the output includes a first number cluster adjacent to a keyword context, converting said first number cluster to a first format based upon the keyword context;

wherein the output includes a second number cluster not adjacent to the keyword, converting said second number cluster to a second format based upon a different set of rules;

12. The method according to claim 11, wherein the second set of rewrite rules is based upon the output of the first set of rewrite rules and the transformed number clusters.

13. The method according to claim 12, further comprising the step of performing at least a second rule interpretation performed to transform text that matches a second set of predetermined patterns.

14. The method according to claim 13, further comprising the step of converting punctuation tokens into symbols that cling to adjacent words.

15. The method according to claim 14, where the adjacent word is capitalized.

16. The method according to claim 14, further comprising the step of removing text used according to a set of speech recognition engine rules.

17. The method according to claim 14, further comprising the step of removing text used according to a set of post processor rules.

18. The method according to claim 16, further comprising the step of writing a file containing post-processed output to a target location.

19. The method according to claim 18, where the target location is a buffer.

20. The method according to claim 18, where the target location is a file.