Focused language models for improved speech input of structured documents

• US 6,901,364 B2
• Filed: 09/13/2001
• Issued: 05/31/2005
• Est. Priority Date: 09/13/2001
• Status: Expired due to Term

• Current Assignee: Panasonic Intellectual Property Corporation of America (Panasonic Holdings Corporation)
• Original Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Inventors: Nguyen, Patrick, Rigazio, Luca, Junqua, Jean-claude
• Primary Examiner(s): Smits, Talivaldis Ivars
• Assistant Examiner(s): Sked, Matthew J
• Application Number:
  US09/951,093
• Publication Number:
  US 20030050778A1
• Time in Patent Office: 1,356 Days
• Field of Search: 704/235, 704/251, 704/275
• US Class Current: 704/235
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

An e-mail message process is provided for use with a personal digital assistant which allows for the use of input speech messaging which is converted to text using a focused language model which is downloaded by a cellular phone connection to an Internet server which provides the focused language model based upon a topic for the intended e-mail message. The text that is generated from the input speech method can be summarized by the e-mail message processor and can be edited by the user. The generated e-mail message can then be transmitted again via cellular connection to an Internet e-mail server for transmitting the e-mail message to a recipient.

Claims

1. A speech recognition processor for processing input speech and converting to text, comprising:

topic determination means for determining a topic of the input speech;

register determination means for determining a register of an outgoing message based on a user-specified register, wherein said register determining means is one or more of a keypad user interface device, a touch-based user interface device, and a speech recognition interface device that allows a user to input said register;

speech input means for allowing a user to input a speech message;

a language model retrieval section for retrieving a focused language model based upon said topic and said register;

a speech recognition module which uses said retrieved focused language model to convert said speech message to text; and

a display section for displaying said text.

2. The speech recognition processor according to claim 1, wherein said topic determination means is a keypad user interface device that allows a user to input said topic.

3. The speech recognition processor according to claim 1, wherein said topic determination means is a speech recognition interface device that allows a user to verbally input said topic.

4. The speech recognition processor according to claim 1, wherein said topic determination means derives said topic from a prestored text message.

5. The speech recognition processor according to claim 1, wherein said language model retrieval section accesses a server via an internet connection to retrieve said language model.

6. The speech recognition processor according to claim 5, wherein said language model retrieval section accesses said server via a wireless connection.

7. The speech recognition processor according to claim 1, further comprising a text summarizing section for summarizing said text.

8. The speech recognition processor according to claim 1, wherein said display section uses an e-mail template for displaying said text.

9. The speech recognition processor of claim 1, further comprising a language model extraction processor receiving information concerning an extracted register attribute and tailoring a language model according to the extracted register attribute.

10. A personal digital computer device, comprising:

a housing including a display screen and an input keypad disposed on an outer surface thereof;

a microphone unit disposed in said housing;

a transmitter/receiver device disposed in said housing;

a processor for processing input speech, including topic determining means for determining a topic of the input speech, register determination means for determining a register of an outgoing message based on a user-specified register, speech input means for allowing a user to input a voice message via said microphone, a language model retrieval section adapted for accessing an internet server via said transmitter/receiver device for retrieving a language model from the internet server based upon said topic and said register, a speech recognition module which uses said retrieved language model to convert said speech message to text, and a display section for displaying said text on said display screen,

wherein said register determining means is one or more of a keypad user interface device, a touch-based user interface device, and a speech recognition interface device that allows a user to input said register.

11. The personal digital computer device according to claim 10, wherein said topic determining means is at least one of a keypad user interface device that allows a user to input said topic.

12. The personal digital computer device according to claim 10, wherein said topic determining means is a speech recognition interface device that allows a user to input said topic.

13. The personal digital computer device according to claim 10, wherein said topic determining means derives said topic from a prestored text message.

14. The personal digital computer device according to claim 10, wherein said language model retrieval section accesses said server via a wireless connection.

15. The personal digital computer device according to claim 10, wherein said processor includes a text summarizing section for summarizing said text.

16. The personal digital computer device according to claim 10, wherein said display section uses an e-mail template for displaying said text.

17. The personal digital computer device of claim 10, further comprising a language model extraction processor receiving information concerning an extracted register attribute and tailoring a language model according to the extracted register attribute.

18. A personal computer implemented speech recognition e-mail processor, comprising:

topic determining means for determining a topic of an outgoing e-mail message;

register determination means for determining a register of an outgoing message based on a user-specified register, wherein said register determining means is one or more of a keypad user interface device, a touch-based user interface device, and a speech recognition interface device that allows a user to input said register;

speech input means for allowing a user to input a voice message;

a language model retrieval section for retrieving a language model based upon said topic and said register;

a speech recognition section which uses said retrieved language model to convert said voice message to text;

a display section for displaying said text in an e-mail template; and

a transmission section for transmitting said e-mail template via a cellular-internet connection.

19. The personal computer implemented speech recognition e-mail processor according to claim 18, wherein said topic determining means is a keypad user interface device that allows a user to input said topic.

20. The personal computer implemented speech recognition e-mail processor according to claim 18, wherein said topic determining means is a speech recognition interface device that allows a user to input said topic.

21. The personal computer implemented speech recognition e-mail processor according to claim 18, wherein said topic determining means derives said topic from a received e-mail message.

22. The personal computer implemented speech recognition e-mail processor according to claim 18, wherein said language model retrieval section accesses said server via a cellular phone connection.

23. The personal computer implemented speech recognition e-mail processor according to claim 18, wherein said processor includes a text summarizing section for summarizing said text.

24. The personal computer implemented speech recognition e-mail processor of claim 18, further comprising a language model extraction processor receiving information concerning an extracted register attribute and tailoring a language model according to the extracted register attribute.

25. A personal computer implemented speech recognition e-mail processor, comprising:

register determining means for determining a register of an outgoing e-mail message based on a user-specified register;

register inferred for an outgoing message replying to a received message based on one or more of;

(a) metadata describing how the received message was formatted, and (b) forms of address present in the received message;

speech input means for allowing a user to input a speech message;

a language model retrieval section for retrieving a language model based upon said register;

a speech recognition section which uses said retrieved language model to convert said speech message to text;

a display section for displaying said text in an e-mail template; and

a transmission section for transmitting said e-mail template via a cellular-internet connection,

wherein said register determining means is one or more of a keypad user interface device and a touch-based user interface device that allows a user to input said register.

26. The personal computer implemented speech recognition e-mail processor according to claim 25, wherein said language model retrieval section accesses said server via a cellular phone connection.

27. The personal computer implemented speech recognition e-mail processor according claim 25, wherein said processor includes a text summarizing section for summarizing said text.

28. The personal computer implemented speech recognition e-mail processor of claim 25, further comprising a language model extraction processor receiving information concerning an extracted register attribute and tailoring a language model according to the extracted register attribute.

29. A personal computer implemented speech recognition e-mail processor, comprising:

register determining means for determining a register of an outgoing e-mail message based on a user-specified register;

register inferred for an outgoing message replying to a received message based on one or more of;

(a) metadata describing how the received message was formatted, and (b) forms of address present in the received message;

speech input means for allowing a user to input a speech message;

a language model retrieval section for retrieving a language model based upon said register;

a speech recognition section which uses said retrieved language model to convert said speech message to text;

a display section for displaying said text in an e-mail template; and

a transmission section for transmitting said e-mail template via a cellular-internet connection,

wherein said register determining means is a speech recognition interface device that allows a user to input said register.

30. The personal computer implemented speech recognition e-mail processor according to claim 29, wherein said language model retrieval section accesses said server via a cellular phone connection.

31. The personal computer implemented speech recognition e-mail processor according to claim 29, wherein said processor includes a text summarizing section for summarizing said text.

32. The personal computer implemented speech recognition e-mail processor of claim 29, further comprising a language model extraction processor receiving information concerning an extracted register attribute and tailoring a language model according to the extracted register attribute.