Intelligent recognition

• US 5,666,400 A
• Filed: 07/07/1994
• Issued: 09/09/1997
• Est. Priority Date: 07/07/1994
• Status: Expired due to Term

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Bell Atlantic Network Services, Inc. (Verizon Communications Inc.)
• Inventors: Meador, Frank, Curry, James, McAllister, Alex
• Primary Examiner(s): Bost, Dwayne
• Assistant Examiner(s): PRESSON, THOMAS
• Application Number:
  US08/271,887
• Time in Patent Office: 1,160 Days
• Field of Search: 379/201, 379/207, 379/88, 379/67, 379/189, 379/188, 379/196, 379/197, 379/198, 379/199, 340/825.3, 340/825.31, 340/825.33, 340/825.34
• US Class Current: 379/88.01
• CPC Class Codes: G10L 15/25 using position of the lips, movement of the lips or face analysis
• CPC Class Codes: G10L 2015/228 of application context
• CPC Class Codes: H04Q 3/0029 Provisions for intelligent networking

Abstract

In a switching system for connecting a call between a calling station and a called station, a system and method of voice recognition using a concentrated or distributed multiplicity of voice recognition and other resources with a facility for selecting an initial resource on the calling station going off-hook by accessing a demographic database and a biometric database using common channel signaling and selecting a prompt to be delivered to the caller from a multiplicity of preselected prompts and reacting to a response by the caller with further addressing of database information to continue to select from said multiplicity of resources the most appropriate resource or resources in reaction to caller utterances. According to another feature the selection of resources is aided by optical means at the calling station delivering information regarding characteristics of the caller including lip movement to permit lip reading.

Claims

1. In a switching system for connecting a calling station to a called station a method of establishing completion of said connection comprising the steps of:

responsive to said calling station going off-hook, establishing the identity of said calling station and using information related to the established identity to address a data base;

responsive to the caller using the handset at said calling station sensing biometric information regarding the caller and using said biometric information to address a data base;

selecting from a plurality of speech recognition resources a resource indicated by at least one of said identity and biometric information;

establishing connection between said off-hook station and a voice platform;

.

inputting a spoken command from a caller at said calling station to said voice platform;

outputting from said plurality of speech recognition resources an output signal representing said spoken command received by said voice platform for use in establishing completion of said connection;

responsive to said spoken command sensing additional biometric information regarding the caller and using said additional biometric information to address a data base; and

outputting from said plurality of speech recognition resources an output signal representing said spoken command;

wherein said output signal is produced by at least one of said selected resources based on an evaluation of at least all sensed biometric information.

2. A method according to claim 1 including the steps of:

deriving from the information from the data base identifying the calling station demographic information; and

said output signal being produced by at least one of said selected resources based on evaluation of said demographic and said biometric information.

3. A method according to claim 1 wherein said biometric information includes information of multiple biometric types.

4. A method according to claim 3 wherein said biometric information includes information resulting from a breath test.

5. A method according to claim 4 wherein said biometric information also includes information resulting from cardiac sensing.

6. A method according to claim 3 wherein said biometric information includes information resulting from cardiac sensing.

7. A method according to claim 3 wherein said biometric information includes information resulting from a plurality of breath tests.

8. A method according to claim 7 wherein said breath tests include tests both before and after connection.

9. In a switching system for connecting a calling station to a called station a method of establishing completion of said connection comprising the steps of:

responsive to said calling station going off-hook, establishing the identity of said calling station and using information related to the established identity to address a data base;

responsive to the caller using the handset at said calling station sensing biometric information regarding the caller and using said biometric information to address a data base;

selecting from a plurality of speech recognition resources a resource indicated by at least one of said identity and biometric information;

establishing connection between said off-hook station and a voice platform;

.

inputting a spoken command from a caller at said calling station to said voice platform; and

outputting from said plurality of speech recognition resources an output signal representing said spoken command received by said voice platform for use in establishing completion of said connection;

wherein said biometric sensing is accomplished at least in part by the caller grasping said handset;

selecting from said plurality of speech recognition resources a second resource responsive to said output signal;

outputting from said plurality of speech recognition resources a second output signal;

inputting a second spoken command from said caller at said calling station to said second resource; and

outputting from said plurality of speech recognition resources a third output signal responsive to the second spoken command.

10. A method according to claim 9 including the steps of:

responsive to said second output signal selecting an audio request from a plurality of preestablished audio requests and outputting to said caller said selected audio request requesting said second command; and

said caller inputting said second spoken command responsive to said selected audio request.

11. A method according to claim 10 including the steps of:

requesting in said second audio request a spoken command of a type to elicit a response to provide deep breath sensing for an active sensing apparatus.

12. A method according to claim 10 wherein said selection of audio requests is based at least in part on biometric information.

13. A method according to claim 12 wherein said biometric information is based on sensing at said handset other than oral sensing.

14. A method according to claim 13 wherein said sensing at said handset other than oral sensing is at least partially derived by electrodes associated with said handset and contacted by the caller.

15. In a Public Switched Telephone Network (PSTN) which includes Common Channel Signaling/CCS) and a Signal Control Point (SCP), a method of completing a call from a calling station to a called station comprising the steps of:

responsive to said calling station going off-hook, establishing the identity of said calling station through said CCS pursuant to the functioning of said PSTN;

using said identity of said calling station to address a data base associated with said SCP;

deriving from said data base information relating to demographics of said calling station;

sensing from said calling station biometric information relating to said caller;

selecting from a plurality of recognition resources a resource indicated by at least one of said demographic information and said biometric information;

establishing a connection to said off-hook station;

inputting a spoken command from a caller at said calling station to said selected resource;

sensing from said calling station additional biometric information relating to said caller and generating a signal representative thereof;

outputting to said caller an audio signal selected from a plurality of preestablished audio signals based on at least one of (a) audio sensing of said spoken command and (b) biometric sensing;

inputting a second spoken command from said caller responsive to said audio signal outputted to said caller;

outputting from said plurality of resources a second output signal responsive to said second spoken command;

determining the degree of traffic through said plurality of resources;

comparing the determined degree of traffic to a predetermined traffic load; and

responsive to said determined degree of traffic being below said predetermined load, inputting at least one of said spoken commands to a plurality of said resources in parallel.

16. In a switching system for connecting a calling station to a called station a method of establishing completion of said connection comprising the steps of:

responsive to said calling station going off-hook, establishing the identity of said calling station and using information related to the established identity to address a data base;

responsive to the caller using the handset at said calling station sensing biometric information regarding the caller and using said biometric information to address a data base;

selecting from a plurality of speech recognition resources a resource indicated by at least one of said identity and biometric information;

establishing connection between said off-hook station and a voice platform;

inputting a spoken command from a caller at said calling station to said voice platform; and

outputting from said plurality of speech recognition resources an output signal representing said spoken command received by said voice platform for use in establishing completion of said connection;

wherein said selecting step comprises the step of selecting said resource in response to said identity information and said biometric information.

17. The method of claim 16 including the steps of:

deriving demographic information from the information from the data base identifying the calling station;

said output signal being produced by at least one of said selected resources based on evaluation of said demographic and said biometric information.

18. A method according to claim 16 wherein said biometric information includes information of multiple biometric types.

19. A method according to claim 18 wherein said biometric information includes information resulting from a breath test.

20. A method according to claim 18 wherein said biometric information includes information resulting from cardiac sensing.

21. In a Public Switched Telephone Network (PSTN) which includes Common Channel Signaling (CCS) and a Signal Control Point (SCP), a method of completing a call from a calling station to a called station comprising the steps of:

responsive to said calling station going off-hook, establishing the identity of said calling station through said CCS pursuant to the functioning of said PSTN;

using said identity of said calling station to address a data base associated with said SCP;

deriving from said data base information relating to demographics of said calling station;

sensing from said calling station biometric information relating to said caller;

selecting from a plurality of recognition resources a resource indicated by at least one of said demographic information and said biometric information;

establishing a connection to said off-hook station;

inputting a spoken command from a caller at said calling station to said selected resource;

sensing from said calling station additional biometric information relating to said caller and generating a signal representative thereof;

outputting to said caller an audio signal selected from a plurality of preestablished audio signals based on at least one of (a) audio sensing of said spoken command and (b) biometric sensing;

inputting a second spoken command from said caller responsive to said audio signal outputted to said caller; and

outputting from said plurality of resources a second output signal responsive to said second spoken command;

wherein said selecting step comprises the step of selecting said resource in response to said demographic information and said biometric information.

22. In a switching system for connecting a calling station to a called station a method of establishing completion of said connection comprising the steps of:

responsive to said calling station going off-hook, establishing the identity of said calling station and using information related to the established identity to address a data base;

responsive to the caller using the handset at said calling station sensing biometric information regarding the caller and using said biometric information to address a data base;

selecting from a plurality of speech recognition resources a resource indicated by at least one of said identity and biometric information;

establishing connection between said off-hook station and a voice platform;

inputting a spoken command from a caller at said calling station to said voice platform;

outputting from said plurality of speech recognition resources an output signal representing said spoken command received by said voice platform for use in establishing completion of said connection;

wherein said selecting step comprises the step of;

tentatively selecting a primary set of said speech recognition resources to determine a tentative recognition result;

finally selecting at least one resource in response to said tentative recognition result for recognition of said spoken command.

23. In a switching system for connecting a calling station to a called station a method of establishing completion of said connection comprising the steps of:

responsive to the caller using the handset at said calling station and going off-hook sensing biometric information regarding the caller using common channel signaling and using said information to address a data base;

selecting from a plurality of speech recognition resources a resource based on said information;

establishing connection to said off-hook station;

inputting a spoken command from a caller at said calling station to said selected resource; and

outputting from said plurality of speech recognition resources an output signal representing said spoken command;

wherein said selecting step comprises the steps of;

tentatively selecting a primary set of said speech recognition resources to determine a tentative recognition result;

finally selecting said at least one resource in response to said tentative recognition result for recognition of said spoken command.

24. A method for recognizing a spoken command from a calling station to complete a call connection in a public switched telephone network, the method comprising the steps of:

establishing an identity of the calling station in response to a detected off-hook condition of the calling station and accessing a database for profile information in response to the established identity;

sensing biometric information of a caller using said calling station in response to the detected off-hook condition;

selecting at least one speech recognition resource from a plurality of speech recognition resources in response to at least one of said profile information and said biometric information;

initiating an interface session between the caller and a voice platform;

receiving the spoken command during said interface session and providing the spoken command from the voice platform to the at least one selected speech recognition resource; and

outputting from the at least one selected speech recognition resource a command signal representing the spoken command;

wherein said selecting step comprises the steps of;

first processing a speech sample received during said interface session to obtain said biometric information of the caller;

second processing said speech sample to obtain language characteristics of said caller;

selecting the at least one selected speech recognition resource in response to at least one of said profile information, said biometric information and said language characteristics.

25. A method as in claim 24, wherein said step of first processing the speech sample comprises the step of determining at least a gender of said caller as said biometric information.

26. A method as in claim 24, wherein said step of second processing the speech sample comprises the step of accessing a database storing predetermined words indicative of behavioral status to determine an emotional state of said caller.

27. A method for recognizing a spoken command from a calling station to complete a call connection in a public switched telephone network, the method comprising the steps of:

establishing an identity of the calling station in response to a detected off-hook condition of the calling station and accessing a database for profile information in response to the established identity;

sensing biometric information of a caller using said calling station in response to the detected off-hook condition;

selecting at least one speech recognition resource from a plurality of speech recognition resources in response to at least one of said profile information and said biometric information;

initiating an interface session between the caller and a voice platform;

receiving the spoken command during said interface session and providing the spoken command from the voice platform to the at least one selected speech recognition resource; and

outputting from the at least one selected speech recognition resource a command signal representing the spoken command;

wherein said biometric information sensing step comprises the step of obtaining an optical representation of said caller;

wherein the step of obtaining an optical representation of said caller includes the step of performing image processing on the optical representation to identify physical characteristics of said caller including at least one of gender and age.

28. A method as in claim 27, wherein the step of obtaining an optical representation of said caller includes the step of processing a video signal of the caller at the calling station to determine mouth movement.

29. A method for recognizing a spoken command from a calling station to complete a call connection in a public switched telephone network, the method comprising the steps of:

establishing an identity of the calling station in response to a detected off-hook condition of the calling station and accessing a database for profile information in response to the established identity;

sensing biometric information of a caller using said calling station in response to the detected off-hook condition;

selecting at least one speech recognition resource from a plurality of speech recognition resources in response to at least one of said profile information and said biometric information;

initiating an interface session between the caller and a voice platform;

receiving the spoken command during said interface session and providing the spoken command from the voice platform to the at least one selected speech recognition resource; and

outputting from the at least one selected speech recognition resource a command signal representing the spoken command;

wherein said selecting step comprises the steps of;

tentatively selecting a primary set of said speech recognition resources to determine a tentative recognition result;

finally selecting said at least one speech recognition resource in response to said tentative recognition result for recognition of said spoken command.