Smart correction of dictated speech

• US 6,418,410 B1
• Filed: 09/27/1999
• Issued: 07/09/2002
• Est. Priority Date: 09/27/1999
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Ortega, Kerry A., Nassiff, Amado
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Nolan, Daniel A.
• Application Number:
  US09/406,661
• Time in Patent Office: 1,016 Days
• Field of Search: 704/231, 704/243, 704/244, 704/250, 704/251, 704/235, 704/270, 704/257
• US Class Current: 704/251
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 2015/0635 updating or merging of old and new templates; Mean values; Weighting

Abstract

In a speech recognition system, a method and system for updating a language model during a correction session can include automatically comparing dictated text to replacement text, determining if the replacement text is on an alternative word list if the comparison is close enough to indicate that the replacement text represents correction of a mis-recognition error rather than an edit, and updating the language model without user interaction if the replacement text is on the alternative word list. If the replacement text is not on the alternative word list, a comparison is made between dictated word digital information and replacement word digital information, and the language model is updated if the digital comparison is close enough to indicate that the replacement text represents correction of a mis-recognition error rather than an edit.

Claims

1. In a speech recognition system, a method of updating a language model during a correction session, comprising the steps of:

automatically comparing a dictated word to a replacement word;

if said comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, determining if said replacement word is on an alternative word list; and

if said replacement word is on said alternative word list, updating said language model without user interaction.

2. The method of claim 1, wherein said replacement word is generated by one of the group consisting of typing over said dictated word, pasting over said dictated word, and deleting said dictated word and replacing it with said replacement word.

3. The method of claim 1, wherein at least one of the group of said dictated word and said replacement word consists of a plurality of words.

4. In a speech recognition system, a method of updating a language model during a correction session, comprising the steps of:

automatically comparing a dictated word to a replacement word;

if said comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, determining if said replacement word is on an alternative word list; and

if said replacement word is not on said alternative word list, comparing dictated word digital information to replacement word digital information, and if said digital comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, updating said language model.

5. The method of claim 4, further comprising the steps of, prior to said digital comparison step:

converting audio of said dictated word into dictated word digital information;

converting said replacement word into replacement word digital information; and

using said dictated word digital information and said replacement word digital information in said digital comparison step.

6. The method of claim 4, wherein said replacement word is generated by one of the group consisting of typing over said dictated word, pasting over said dictated word, and deleting said dictated word and replacing it with said replacement word.

7. The method of claim 4, wherein at least one of the group of said dictated word and said replacement word consists of a plurality of words.

8. A system for updating a language model during a correction session, comprising:

a means for automatically comparing a dictated word to a replacement word;

if said comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, a means for determining if said replacement word is on an alternative word list; and

if said replacement word is on said alternative word list, a means for updating said language model without user interaction.

9. The system of claim 8, where in said replacement word is generated by one of the group consisting of a means for typing over said dictated word, a means for pasting over said dictated word, and a means for deleting said dictated word and replacing it with said replacement word.

10. The system of claim 8, wherein at least one of the group of said dictated word and said replacement word consists of a plurality of words.

11. A system for updating a language model during a correction session, comprising:

a means for automatically comparing a dictated word to a replacement word;

if said comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, a means for determining if said replacement word is on an alternative word list; and

if said replacement word is not on said alternative word list, a means for comparing dictated word digital information to replacement word digital information, and if said digital comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, a means for updating said language model.

12. The system of claim 11, further comprising:

a means for converting audio of said dictated word into dictated word digital information;

a means for converting said replacement word into replacement word digital information; and

a means for using said dictated word digital information and said replacement word digital information in said digital comparing means.

13. The system of claim 11, wherein said replacement word is generated by one of the group consisting of a means for typing over said dictated word, a means for pasting over said dictated word, and a means for deleting said dictated word and replacing it with said replacement word.

14. The system of claim 11, wherein at least one of the group of said dictated word and said replacement word consists of a plurality of words.

15. A machine readable storage, having stored thereon a computer program having a plurality of code sections executable by a machine for causing the machine to perform the steps of:

automatically comparing a dictated word to a replacement word;

if said comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, determining if said replacement word is on an alternative word list; and

if said replacement word is on said alternative word list, updating said language model without user interaction.

16. The machine readable storage of claim 15, wherein said replacement word is generated by one of the group consisting of typing over said dictated word, pasting over said dictated word, and deleting said dictated word and replacing it with said replacement word.

17. The machine readable storage of claim 15, wherein at least one of the group of said dictated word and said replacement word consists of a plurality of words.

18. A machine readable storage, having stored thereon a computer program having a plurality of code sections executable by a machine for causing the machine to perform the steps of:

automatically comparing a dictated word to a replacement word;

if said comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, determining if said replacement word is on an alternative word list; and

if said replacement word is not on said alternative word list, comparing dictated word digital information to replacement word digital information, and if said digital comparison is close enough, within a predetermined statistical quantity, to indicate that said replacement word represents correction of a misrecognition error rather than an edit, updating said language model.

19. The machine readable storage of claim 18, further comprising the steps of, prior to said acoustic comparison step:

converting audio of said dictated word into dictated word digital information;

converting said replacement word into replacement word digital information; and

using said dictated word digital information and said replacement word digital information in said digital comparison step.

20. The machine readable storage of claim 18, wherein said replacement word is generated by one of the group consisting of typing over said dictated word, pasting over said dictated word, and deleting said dictated word and replacing it with said replacement word.

21. The machine readable storage of claim 18, wherein at least one of the group of said original dictated text and said replacement text consists of a plurality of words.