Message recognition using shared language model

• US 6,904,405 B2
• Filed: 01/28/2002
• Issued: 06/07/2005
• Est. Priority Date: 07/17/1999
• Status: Expired due to Term

• Current Assignee: Buffalo Patents, LLC
• Original Assignee: Optical Research Partners LLC (Intellectual Ventures LLC)
• Inventors: Suominen, Edwin A.
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Armstrong, Angela A.
• Application Number:
  US10/061,052
• Publication Number:
  US 20030055655A1
• Time in Patent Office: 1,226 Days
• Field of Search: 704/231, 704/235, 704/254, 704/257, 704/260, 704/270
• US Class Current: 704/235
• 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

Certain disclosed methods and systems perform multiple different types of message recognition using a shared language model. Message recognition of a first type is performed responsive to a first type of message input (e.g., speech), to provide text data in accordance with both the shared language model and a first model specific to the first type of message recognition (e.g., an acoustic model). Message recognition of a second type is performed responsive to a second type of message input (e.g., handwriting), to provide text data in accordance with both the shared language model and a second model specific to the second type of message recognition (e.g., a model that determines basic units of handwriting conveyed by freehand input). Accuracy of both such message recognizers can be improved by user correction of misrecognition by either one of them. Numerous other methods and systems are also disclosed.

Claims

1. A system for generating text responsive to message input of a first type and a second type, the system comprising:

(a) a unified message model including;

(1) a shared language model;

(2) a first model specific to a first type of message recognition; and

(3) a second model specific to a second type of message recognition;

(b) a first message recognizer, responsive to the first type of message input to provide text data in accordance with both the shared language model and the first model; and

(c) a second message recognizer, responsive to the second type of message input to provide text data in accordance with both the shared language model and the second model;

wherein

(d) the shared language model is trainable responsive to user correction of misrecognition by either of the first and second message recognizers, thereby improving accuracy of each of the first and second message recognizers.

2. The system of claim 1 wherein:

(a) the first type of message recognition is speech recognition, the first type of message input is voice input, and the first model is an acoustic model; and

(b) the second type of message recognition is handwriting recognition, the second type of message input is freehand input, and the second model is a handwriting model.

3. The system of claim 2 further comprising a stylus, wherein the shared language model is trainable responsive to user correction of misrecognition by the first message recognizer by (a) movement of the stylus across visual indicia of dictated text that has been erroneously generated to specify a selection from the text, and (b) correction of the selection.

4. The system of claim 3 wherein:

(a) the stylus includes a pointing interface and a microphone;

(b) the system further comprises a tablet responsive to the pointing interface;

(c) the voice input is from the microphone; and

(d) the freehand input is from the pointing interface being manually manipulated on the tablet.

5. The system of claim 4 wherein the visual indicia of dictated text appears on the tablet.

6. The system of claim 1 wherein the shared language model comprises a syntactic model and a semantic model.

7. A method for performing message recognition with a shared language model, the method comprising:

(a) performing message recognition of a first type, responsive to a first type of message input, to provide text data in accordance with both the shared language model and a first model specific to the first type of message recognition;

(b) performing message recognition of a second type, responsive to a second type of message input, to provide text data in accordance with both the shared language model and a second model specific to the second type of message recognition; and

(c) training the shared language model responsive to user correction of error in message recognition of either of the first and second types, thereby improving accuracy of each of the first arid second types of message recognition.

8. The method of claim 7 wherein:

(a) performing the first type of message recognition comprises performing speech recognition responsive to voice input in accordance with the shared language model and a first model that is an acoustic model; and

(b) performing the second type of message recognition comprises performing handwriting recognition responsive to freehand input in accordance with the shared language model and a second model that is a handwriting model.

9. The method of claim 8 wherein training the shared language model includes (a) moving a stylus across visual indicia of dictated text that has been erroneously generated to specify a selection from the text, and (b) correcting the selection.

10. The method of claim 9 wherein:

(a) performing speech recognition responsive to voice input includes responding to voice input from a microphone in the stylus; and

(b) performing handwriting recognition responsive to freehand input includes responding to manual manipulation of the stylus on a tablet.

11. The method of claim 10 wherein the visual indicia of dictated text appears on the tablet.

12. The method of claim 7 wherein performing message recognition in accordance with the shared language model comprises performing message recognition in accordance with a syntactic model and a semantic model.

13. A system for performing message recognition with a trainable shared language model, the system comprising:

(a) means for performing message recognition of a first type, responsive to a first type of message input, to provide text data in accordance with both the shared language model and a first model specific to the first type of message recognition; and

(b) means for performing message recognition of a second type, responsive to a second type of message input, to provide text data in accordance with both the shared language model and a second model specific to the second type of message recognition.

14. The system of claim 13 wherein:

(a) the first type of message recognition is speech recognition, the first type of message input is voice input, and the first model is an acoustic model; and

(b) the second type of message recognition is handwriting recognition, the second type of message input is freehand input,. and the second model is a handwriting model.

15. The system of claim 13 wherein the shared language model comprises a syntactic model and a semantic model.

16. The system of claim 13 further comprising means for training the shared language model responsive to user correction of error in message recognition of either of the first and second types, thereby improving accuracy of each of the first and second types of message recognition.

17. The system of claim 16 wherein the means for training the shared language model includes means for (a) specifying a selection from visual indicia of dictated text that has been erroneously generated, and (b) correcting the selection.