Method and apparatus for fast voice selection of error words in dictated text

• US 5,909,667 A
• Filed: 03/05/1997
• Issued: 06/01/1999
• Est. Priority Date: 03/05/1997
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Coe, Kris, Leontiades, Kyriakos
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): SAX, ROBERT L
• Application Number:
  US08/812,387
• Time in Patent Office: 818 Days
• Field of Search: 704/235, 704/251, 704/275
• US Class Current: 704/275
• CPC Class Codes: G06F 3/167 Audio in a user interface, e.g. using voice commands for navigating, audio feedback
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

A computer voice operated dictations system having a text editing mode provided for editing dictated text. The text editing mode navigates a user through the dictated text by highlighting a correction window comprised of a plurality of words comprising a portion of the text. The correction window is comprised of "m" words and the value of m is preferably determined, based upon the voice recognition capabilities of a particular voice recognition engine associated with the dictation system. The greater the ability of the speech recognition engine to recognize and distinguish spoken words, the larger the value of m. A correction vocabulary is defined which consists of the plurality of words which are contained within the correction window and a plurality of correction commands. The voice recognition computer dictation system is configured for processing a spoken utterance which corresponds to an error word contained within the plurality of words highlighted in the correction window is processed. When the error word is spoken by a user it uniquely identifies the word which is to be edited. Thereafter, the error word can be immediately corrected by the user articulating a further voice command.

Claims

1. In a computer system (20) having an operating system (24), a memory (27), and processing means (26, 28) for receiving spoken words and converting said spoken words to words for storage in said memory (27) and for display on a visual display (32), a method for voice-activated correction of text, said method comprising:

(a) storing in said memory (27) a plurality of words (60) corresponding to said spoken words which are enabled for correction by voice activation, alternate word lists containing alternate words which are also likely matches for each spoken word, and a plurality of correction commands;

(b) displaying one or more of the words (60) enabled for correction in a first demarcated correction region (130) on the visual display (32) in response to one of said plurality of correction commands being spoken by a user, said demarcated correction region identifying words which are available for a user to say and correct;

(c) receiving and identifying in the processing means (26,

     28), an error word (132) spoken by a user and matching said spoken error word to a word which exists in said demarcated correction region;

(d) operating said processing means (26,

     28) to retrieve from said memory (27) and to display on said visual display (32), one or more alternate words (136) from said alternate word list for said error word as potential replacements for said error word (132);

(e) receiving and identifying in the processing means (26,

     28) a second one of said plurality of correction commands, in response to said second one of said plurality of correction commands (136,

     138) being spoken by a user; and

wherein the processing means (26,

     28) is responsive to said second one of said plurality of correction commands to correct the error word on the visual display.

2. The method according to claim 1, further comprising the steps of

(e) displaying a subsequent demarcated correction region, said subsequent demarcated correction region beginning with a word following the error word contained in the first demarcated correction region (130), and

(f) repeating steps (c)-(e) for the subsequent demarcated correction region until the correction of text is completed.

3. The method according to claim 1, wherein said first demarcated correction region (130) begins at a position where a cursor is initially displayed on said user interface screen when the voice activated correction process is initiated.

4. The method according to claim 1, wherein said demarcated correction region (130) begins at a line of text where a cursor is currently displayed on said user interface screen.

5. The method according to claim 1 wherein the one or more words enabled for correction may be user selected by correction commands which reposition said demarcated correction region (130).

6. The method according to claim 1 wherein said one or more words enabled for correction are demarcated by drawing a thin line border around the text.

7. The method according to claim 1 wherein said one or more words enabled for correction are demarcated by drawing a thin line adjacent to said words.

8. The method according to claim 1 wherein said one or more words enabled for correction are demarcated by displaying said words in a color which is different from the remainder of the words enabled for correction.

9. The method according to claim 1 wherein said text is demarcated by underlining.

10. The method according to claim 1 wherein said first demarcated correction region is comprised of m words and the value of m is statistically determined, based upon the voice recognition capabilities of the voice recognition engine, to be large enough so as to include at least one occurrence of a word which requires correction in the case of previously unedited text which has been dictated.

11. The method according to claim 10 wherein the value of m is statistically determined, based upon the voice recognition capabilities of the voice recognition engine, to avoid frequent misidentification of a word which requires correction.

12. The method according to claim 11 wherein the value of m is equal to between about 10 and 50.

13. The method according to claim 1 wherein said processing means is comprised of a speech recognition engine (26) and a speech text processor (28).

14. The method according to claim 13 wherein said processing means is further comprised of a voice navigator application (22) for coordinating the operation of said speech recognition engine (26), the speech text processor (28) and any other application programs executing on said computer system.

15. The method according to claim 13 wherein said speech recognition engine (26) converts a spoken word to text, and identifies said spoken word to said speech text processor in a word identification message which includes a tag value.

16. The method according to claim 13 further comprising the steps of:

assigning a unique tag value to each word converted to text by said speech recognition engine and providing said tag value to said speech text processor.

17. The method according to claim 16 further comprising the step of storing in memory (27) a word list for each of said speech recognition engine and said speech text processor, each of said word lists being comprised of a list of words (60) converted to text by said speech recognition engine ((26) and the tag values (62) assigned to each word.

18. The method according to claim 17 further comprising the steps of storing in memory (27) as part of said word list for said speech recognition engine, an alternate word list (64) comprised of words identified by said speech recognition engine (26) as possible alternative matches to each of said spoken words, and storing in memory as part of said word list for said speech text processor a word index number (66) for each word, said word index number (66) uniquely identifying a position of each word in a document.

19. The method according to claim 18 wherein said alternate words (64) are obtained by said speech text processor (28) for presentation on said visual display (32) by the further steps of:

identifying said error word by matching a word spoken by a user to one of said words enabled for correction in said demarcated correction region (130);

retrieving from the word list of said speech text processor a tag value (62) for the error word (132); and

retrieving from said word list of said speech recognition engine the alternate word list (64) for the word uniquely identified by said tag value (62).

20. In a computer system (20) having an operating system (24), a computer memory (27), and a user interface display (32), an apparatus for voice-activated correction of text, said apparatus comprising:

(a) means for storing in said memory (27) a plurality of words (60) corresponding to words spoken by a user which are enabled for correction by voice activation, alternate word lists containing alternate words which are also likely matches for each spoken word, and a plurality of voice activated correction commands;

(b) means for displaying one or more of the words (60) enabled for correction in a first demarcated correction region (130) on the visual display (32), said demarcated correction region identifying words which are available for a user to say and correct;

(c) means for receiving and identifying an error word (132) to be corrected when said error word is spoken by a user, said error word (132) being contained within said demarcated correction region, and matching said word to a word which exists in said demarcated correction region;

(d) means for retrieving from said memory and displaying on said visual display (32) one or more alternate words (136) from said alternate word list for said error word as potential replacements for said error word;

(e) means responsive to one of said plurality of correction commands being spoken by a user, for correcting the error word on the visual display.

21. The apparatus according to claim 20 wherein the means for displaying said one or more of the words enabled for correction in a first demarcated correction region (130) on the visual display (32) is configured to display a subsequent demarcated correction region after said error word (132) has been corrected.

22. The apparatus according to claim 21 wherein said demarcated correction region is comprised of m words.

23. The apparatus according to claim 21 wherein said system is configured to permit a subsequent error word in said subsequent demarcated correction region to be corrected.

24. In a computer system having an operating system, a memory, and processing means for receiving spoken words and converting spoken words to text words for storage in memory and display on a visual display, an apparatus for voice-activated correction of text, said apparatus comprising:

(a) text processing means for

(1) selecting a first set of correction words enabled for voice activated correction from the text words stored in said memory,

(2) for enabling a first group of voice activated correction commands, and

(3) for displaying the first set of correction words in a first demarcated correction region on the visual display;

(b) means for receiving and storing in said memory (27) an error word (132), when said error word (132) is spoken by a user, and for communicating said error word (132) to said text processing means;

(c) means for matching said error word (132) with one of said correction words;

(d) means for displaying a list of alternate words (64) on the visual display (32) of the computer system (20) in response to said error word (132) being spoken, said alternate word list created for each word spoken during a dictation mode, said alternate word list containing alternate words which are also likely matches for said spoken error word;

(e) means for receiving and storing a spoken correction command (136,

     138) in the computer system to replace said error word with a word contained on said alternate word list;

whereby said computer system is responsive to the spoken correction command to correct the error word on the visual display.

25. The apparatus according to claim 24 further comprising means for selecting a subsequent set of correction words from the text words stored in said memory, said subsequent set of correction words enabled for voice activated correction, and for displaying the subsequent set of correction words in a subsequent demarcated correction region on the visual display for correction.

26. The apparatus according to claim 24 wherein said spoken correction command is a unique character associated with one of said alternate words (64).

27. The apparatus according to claim 25 wherein said subsequent demarcated correction region is comprised of m words and begins with the word following the error word contained in the first demarcated correction region.