Historical database storing relationships of successively spoken words

• US 5,970,448 A
• Filed: 07/23/1993
• Issued: 10/19/1999
• Est. Priority Date: 06/01/1987
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Kurzweil Applied Intelligence, Inc. (Intel Corporation)
• Inventors: Wilson, Brian D., Hume, Christopher N., Dooley, John F., Lerner, James P., Goldhor, Richard S.
• Primary Examiner(s): Knepper, David D.
• Application Number:
  US08/096,686
• Time in Patent Office: 2,279 Days
• Field of Search: 381/43, 381/44, 395/2.44, 395/2.79, 395/2.84, 395/2.86, 395/2.87, 704/235, 704/251, 704/270, 704/275, 704/277, 704/278, 704/255
• US Class Current: 704/235
• CPC Class Codes: G06F 3/16 Sound input; Sound output speech processing G10L
• CPC Class Codes: G10L 15/18 using natural language modelling
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

The text is generated from voice input that divides the processing of each spoken word into a dictation event and a text event. Each dictation event handles the processing of data relating to the input into the system, and each text event deals with the generation of text from the inputted voice signals. In order to easily distinguish the dictation events from each other and text events from each other the system and method creates a data structure for storing certain information relating to each individual event. Such data structures enable the system and method to process both simple spoken words as well as spoken commands and to provide the necessary text generation in response to the spoken words or to execute an appropriate function in response to a command.

Claims

1. A system for generating text in response to a succession of audio signals representing spoken input events provided by a user, said system comprising:

means for comparing each spoken input event with a plurality of tokens representing vocabulary words thereby to identify a plurality of candidate tokens which may correspond to the spoken input event, each candidate being scored as to likelihood of match;

computer implemented means for generating and storing, for each spoken input event, a data record which includes;

the identity of the best matching candidate token;

the identity of the correct candidate; and

data placing the chronology of the data record relative other data records in the database;

thereby to generate a dictation event database useful for improving recognizer accuracy by learning a user'"'"'s speech behavior.

2. A system as set forth in claim 1 wherein each said data record also includes the identity of the candidate tokens.

3. A system as set forth in claim 1 wherein each said data record also includes information defining the state of said comparing means at the time the respective data record was generated.

4. A system as set forth in claim 3 wherein said vocabulary words include text words and command words and wherein one of said command words causes the generation of a number of rubout characters equal to the number of characters in the correct candidate word and a resetting of the state of the comparing means back to the state defined in the respective data record.

5. A system for generating text in response to a succession of audio signals representing spoken input events provided by a user, said system comprising:

means for comparing each spoken input event with a plurality of tokens representing vocabulary words thereby to identify a plurality of candidate words which may correspond to the spoken input event, each candidate being scored as to likelihood of match;

means enabling a user to accept one of said candidate words;

computer implemented means for generating and storing, for each spoken input event, a data record which includes;

the identity of the candidate tokens

the identity of the best matching candidate token

the identity of the word accepted by the user and

data placing the chronology of the respective data record relative other data records;

thereby to generate a dictation event database useful for backing up in the generated text and changing the word accepted by the user.

6. A system as set forth in claim 5 wherein each said data record also includes information defining the state of said comparing means at the time the respective data record was generated.

7. A system as set forth in claim 5 wherein the generated text and said database can be stored thereby facilitating subsequent editing of the generated text by changing the word accepted from the candidate set.

8. A system for generating text in response to a succession of audio signals representing spoken input events provided by a user, said system comprising:

means for comparing each spoken input event with a plurality of tokens representing vocabulary words thereby to identify a plurality of candidate tokens which may correspond to the spoken input event;

means enabling a user to accept one of said candidate tokens;

computer implemented means responsive to an accepted candidate token for generating and storing a data record which includes;

data defining the respective input event

data placing the chronology of the record with respect to other similarly generated records

data defining any hierarchical relationship to other similarly generated records;

thereby to generate a text event database useful in backing up in the generated text and making corrections.

9. A system for generating text in response to a succession of audio signals representing spoken input events provided by a user, said system comprising:

means for comparing each spoken input event with a plurality of tokens representing vocabulary words thereby to identify a plurality of candidate words which may correspond to the spoken input event, each candidate being scored as to likelihood of match;

means enabling a user to accept one of said candidate words;

computer implemented means for generating and storing, for each spoken input event, a dictation event data record which includes;

the identity of the candidate words

the identity of the best matching candidate word

the identity of the word accepted by the user and

data placing the chronology of the respective data record relative other data records;

means responsive to an accepted candidate word for generating a text event data record which includes;

data defining the respective input event

data placing the chronology of the record with respect to other similarly generated records

data defining any hierarchical relationship to other similarly generated records;

thereby to generate a database useful in backing up in the generated text and making corrections.

10. A system as set forth in claim 9 wherein the text event data records also include a field identifying a corresponding dictation event data record.

11. A system for generating text in response to a succession of input events provided by a user, said input events including audio signals representing speech input and manual events representing manual input events, said system comprising:

means for comparing each spoken input event with a plurality of tokens representing vocabulary words thereby to identify a plurality of candidate tokens which may correspond to the spoken input event;

means enabling a user to accept one of said candidate tokens;

computer implemented means responsive to an accepted candidate token or a manual event for generating and storing a data record which includes;

data defining the type of input event

data placing the chronology of the record with respect to other similarly generated records

thereby to generate a text event database useful in backing up in the generated text and making corrections.

12. A system as set forth in claim 11 wherein said manual input includes keyboard operations.

13. A system as set forth in claim 11 wherein said manual input includes operation of a pointing device.

14. A system for generating structured text in response to a succession of audio signals representing spoken input events provided by a user, said system comprising:

a plurality of tokens representing vocabulary words including both text words and command words, at least one of said command words being associated with a text form having fields to be filled in;

means for comparing each spoken input event with at least a preselected group of said tokens thereby to identify candidate words which correspond to the spoken input events;

means enabling a user to accept a candidate word;

computer implemented means responsive to an accepted candidate word for generating and storing a text event data record which includes;

data defining the respective input event

data defining any hierarchical relationship to other similarly generated records, text event data records associated with filling in fields in said text form being inferior to a text event data record associated with said text form;

thereby to generate a text event database useful in subsequent editing of the structured text.

15. A system for generating structured text in response to a succession of audio signals representing spoken input events provided by a user, said system comprising:

a plurality of tokens representing vocabulary words including both text words and command words, at least one of said command words being associated with a text form having fields to be filled in;

means for comparing each spoken input event with at least a preselected group of said tokens thereby to identify candidate words which correspond to the spoken input events;

means enabling a user to accept a candidate word;

computer implemented means for generating and storing, for each spoken input event, a data record which includes;

the identity of the candidate words

the identity of the best matching candidate word and

the identity of the correct candidate;

computer implemented means responsive to an accepted candidate word for generating and storing a text event data record which includes;

data defining the respective input event

data defining any hierarchical relationship to other similarly generated records, text event data records associated with filling in fields in said text form being inferior to a text event data record associated with said text form;

thereby to generate a database useful in subsequent editing of the structured text.

16. A system as set forth in claim 15 wherein the text event data records also include a field identifying a corresponding dictation event data record.