Method and system using a speech recognition system to dictate a body of text in response to an available body of text

• US 6,308,151 B1
• Filed: 05/14/1999
• Issued: 10/23/2001
• Est. Priority Date: 05/14/1999
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Smith, Maria E.
• Primary Examiner(s): Dorvil, Richemond
• Application Number:
  US09/312,724
• Time in Patent Office: 893 Days
• Field of Search: 704/235, 704/260, 704/251, 704/231, 704/255, 704/257, 704/275, 704/256, 704/270, 704/272, 704/277
• US Class Current: 704/235
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• 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 2015/228 of application context

Abstract

A system and method is delineated for dictating a body of text in response to an available body of text. In the preferred embodiment, the available body of text comprises only the textual body in plain format from a received E-mail message, while the responsive body of text preferably comprises a dictated E-mail response. Each word from the selected text of the received E-mail message is compared against the speech recognition system'"'"'s vocabulary to determine whether any words are out-of-vocabulary. Out-of-vocabulary words and their pronunciations are added to the system vocabulary. Similarly, new context information is extracted from the selected text of the received E-mail message, and used to update the system'"'"'s language model. Thereafter, the user more accurately and efficiently dictates the responsive E-mail, and the system removes the updates to the system vocabulary and language model.

Claims

1. In a computer system for speech recognition, a method for dictating a body of text in response to an available body of text comprising the steps of:

retrieving from a memory the available body of text;

identifying out-of-vocabulary words from the available body of text by comparing each word from the available body of text against words in the speech recognition system'"'"'s vocabulary;

updating the system'"'"'s vocabulary to temporarily include the out-of-vocabulary words;

dictating the responsive body of text; and

removing the out-of-vocabulary words from the system'"'"'s vocabulary after dictating the responsive body of text.

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

retrieving pronunciations of the out-of-vocabulary words from a pronunciation database;

generating pronunciations of the out-of-vocabulary words not having pronunciations in the database;

updating the system'"'"'s vocabulary to temporarily include the retrieved and generated pronunciations; and

removing from the system'"'"'s vocabulary the retrieved and generated pronunciations after dictation of the responsive body of text.

3. The method of claim 2 wherein the out-of-vocabulary words, and the retrieved and generated pronunciations are removed from the system'"'"'s vocabulary following an interval beginning after dictation of the responsive body of text and running for a period of time calculable in a predetermined manner.

4. The method according to claim 1 further comprising the steps of:

generating context information from the available body of text;

updating the system'"'"'s language model to temporarily include the context information; and

removing the context information generated from the available body of text after dictation of the responsive body of text.

5. The method of claim 4 wherein the context information updates are removed from the language model following an interval beginning after dictation of the responsive body of text and running for a period of time calculable in a predetermined manner.

6. The method according to claim 1 further comprising the step of determining whether a user'"'"'s spoken command decodes into a command indicative of a situation where the user is responding to an E-mail message.

7. The method of claim 1 wherein the available body of text comprises a received E-mail message'"'"'s body in plain text format.

8. The method of claim 1 wherein the dictated body of text comprises an E-mail message responsive to a received E-mail message.

9. A system for dictating a body of text in response to an available body of text comprising:

means for retrieving from a memory the available body of text;

means for identifying out-of-vocabulary words from the available body of text by comparing each word from the available body of text against words in the speech recognition system'"'"'s vocabulary;

means for updating the system'"'"'s vocabulary to temporarily include the out-of-vocabulary words;

means for dictating the responsive body of text; and

means for removing the out-of-vocabulary words from the system'"'"'s vocabulary after dictating the responsive body of text.

10. The system according to claim 9 further comprising:

means for retrieving pronunciations of the out-of-vocabulary words from a pronunciation database;

means for generating pronunciations of the out-of-vocabulary words not having pronunciations in the database;

means for updating the system'"'"'s vocabulary to temporarily include the retrieved and generated pronunciations; and

means for removing from the system'"'"'s vocabulary the retrieved and generated pronunciations after dictation of the responsive body of text.

11. The system according to claim 9 further comprising:

means for generating context information from the available body of text;

means for updating the system'"'"'s language model to temporarily include the context information; and

means for removing the context information generated from the available body of text after dictation of the responsive body of text.

12. The system according to claim 9 further comprising means for determining whether a user'"'"'s spoken command decodes into a command indicative of a situation where the user is responding to an E-mail message.

13. The system of claim 9 wherein the available body of text comprises a received E-mail message'"'"'s body in plain text format.

14. The system of claim 9 wherein the dictated body of text comprises an E-mail message responsive to a received E-mail message.

15. A machine readable storage medium, 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:

retrieving from a memory an available body of text;

identifying out-of-vocabulary words from the available body of text by comparing each word from the available body of text against words in a speech recognition system'"'"'s vocabulary;

updating the system'"'"'s vocabulary to temporarily include the out-of-vocabulary words;

dictating a responsive body of text; and

removing the out-of-vocabulary words from the system'"'"'s vocabulary after dictating the responsive body of text.

16. The machine readable storage medium of claim 15 further causing the machine to perform the steps of:

retrieving pronunciations of the out-of-vocabulary words from a pronunciation database;

generating pronunciations of the out-of-vocabulary words not having pronunciations in the database;

updating the system'"'"'s vocabulary to temporarily include the retrieved and generated pronunciations; and

removing from the system'"'"'s vocabulary the retrieved and generated pronunciations after dictation of the responsive body of text.

17. The machine readable storage medium of claim 15 further causing the machine to perform the steps of:

generating context information from the available body of text;

updating the system'"'"'s language model to temporarily include the context information; and

removing the context information generated from the available body of text after dictation of the responsive body of text.