Method for dynamic learning of individual voice patterns

• US 8,655,660 B2
• Filed: 02/10/2009
• Issued: 02/18/2014
• Est. Priority Date: 12/11/2008
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Szalai, Zsolt, Bazot, Philippe, Pucci, Bernard, Vitale, Joel
• Primary Examiner(s): Desir, Pierre-Louis
• Assistant Examiner(s): Serrou, Abdelali
• Application Number:
  US12/368,349
• Publication Number:
  US 20100153108A1
• Time in Patent Office: 1,834 Days
• Field of Search: 704/243, 704/246, 704/270, 704/270.1, 704/271, 704/273, 704/260, 704/277, 704/278, 704/E15.004, 704/E15.007, 704/E15.014, 704/E19.007
• US Class Current: 704/260
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: G10L 17/04 Training, enrolment or model building

Abstract

The present invention is a system and method for generating a personal voice font including, monitoring voice segments automatically from phone conversations of a user by a voice learning processor to generate a personalized voice font and delivering the personalized voice font (PVF) to the a server.

Claims

1. A method for generating a personal voice font comprising:

training a voice learning processor to;

having a user speak a name of the user and a set of words to provide for recognition of a voice of the user by the voice learning processor;

automatically gathering a plurality of phone conversations of the user when the voice learning processor recognizes the voice of the user;

filtering out voices that do not correspond to the voice of the user from the plurality of phone conversations of the user;

extracting recognized voice pattern data of the user from the plurality of filtered phone conversations of the user;

generating a level of completion message for extracting the recognized voice pattern data to the user after each of the plurality of phone conversations of the user;

generating an error message to the user when the recognized voice pattern data from one of the plurality of phone conversations of the user is not useful, and ignoring the not useful recognized voice data;

generating a personalized voice font (PVF) when sufficient recognized voice pattern data of the user is extracted from the plurality of phone conversations of the user based on the level of completion that indicates completion of the extracted voice pattern data and generating a temporary PVF when the level of completion indicates incomplete extraction of voice pattern data; and

delivering the (PVF) to a server.

2. The method of claim 1, further comprising:

creating a universal voice identifier through which the PVF can be made available to a receiver.

3. The method of claim 1, wherein the voice learning processor discards unrecognizable voice segments from the plurality of phone conversations of the user.

4. The method of claim 1, wherein the monitoring can be initialized, stopped, resumed or restarted by the user.

5. The method of claim 1, wherein subscription to the voice learning processor is required.

6. The method of claim 1, wherein the user instructs the voice learning processor through numerical key codes on a phone.

7. A system for generating a personal voice font comprising:

training a voice learning processor by having;

a server for automatically monitoring and storing a plurality of phone conversations of a user wherein the user initializes the system by speaking a name of the user and a set of words to provide for recognition of a voice of the user by the voice learning processor;

the voice learning processor configured to filter out voices that do not correspond to the voice of the user from the plurality of phone conversations of the user, to extract recognized voice pattern data of the user from the plurality of filtered phone conversations of the user, to generate a level of completion message to extract the recognized voice pattern data to the user for each of the plurality of phone conversations of the user;

to generate an error message to the user when the recognized voice pattern data from one of the plurality of phone conversations of the user is not useful, and ignoring the not useful recognized voice data, and to generate a personalized voice font (PVF) when sufficient recognized voice pattern data of the user is extracted from the plurality of phone conversations of the user based on the level of completion that indicates completion of the extracted voice pattern data and generating a temporary PVF when the level of completion indicates incomplete extraction of voice pattern data; and

delivering the PVF to the server.

8. The system of claim 7, wherein the voice learning processor creates a universal voice identifier associated with the PVF so that the PVF can be made available to a receiver.

9. The system of claim 7, wherein the voice learning processor discards unrecognizable voice segments from the plurality of phone conversations of the user.

10. The system of claim 7, wherein the server monitoring can be initialized, stopped, resumed or restarted by the user.

11. The system of claim 7, wherein subscription is required.

12. A non-transitory computer readable medium having a computer program product stored thereon that when executed on a computer comprises:

training a voice learning processor by;

automatically monitoring and storing a plurality of phone conversations of a user wherein the user initializes the system by speaking a name of the user and a set of words to provide for recognition of a voice of the user;

filtering out voices that do not correspond to the voice of the user from the plurality of phone conversations of the user,

extracting recognized voice pattern data of the user from the plurality of filtered phone conversations of the user;

generating a level of completion message for extracting the recognized voice pattern data to the user after each of the plurality of phone conversations of the user;

generating an error message to the user when the recognized voice pattern data from one of the plurality of phone conversations of the user is not useful, and ignoring the not useful recognized voice data;

generating a personalized voice font (PVF) when sufficient recognized voice pattern data of the user is extracted from the plurality of phone conversations of the user based on the level of completion that indicates completion of the extracted voice pattern data and generating a temporary PVF when the level of completion indicates incomplete extraction of voice pattern data; and

delivering the (PVF) to a server.

13. The non-transitory computer readable medium of claim 12, further comprising:

creating a universal voice identifier through which the PVF can be made available to a receiver.

14. The non-transitory computer readable medium of claim 12, further comprising:

discarding unrecognizable voice segments from the plurality of phone conversations of the user.

15. A method for deploying a system for generating a personal voice font comprising:

training a voice learning processor by;

providing a computer infrastructure being operable to;

have a user speak a name of the user and a set of words to provide for recognition of a voice of the user by the voice learning processor;

automatically gather a plurality of phone conversations of the user when the voice learning processor recognizes the voice of the user;

filter out voices that do not correspond to the voice of the user from the plurality of phone conversations of the user;

extract recognized voice pattern data of the user from the plurality of filtered phone conversations of the user;

generate a level of completion message for extracting the recognized voice pattern data to the user after each of the plurality of phone conversations of the user;

generate an error message to the user when the recognized voice pattern data from one of the plurality of phone conversations of the user is not useful, and ignoring the not useful recognized voice data;

generate a personalized voice font (PVF) when sufficient recognized voice pattern data of the user is extracted from the plurality of phone conversations of the user based on the level of completion that indicates completion of the extracted voice pattern data and generating a temporary PVF when the level of completion indicates incomplete extraction of voice pattern data; and

deliver the (PVF) to a server.