Voice dialing using speech models generated from text and/or speech

• US 6,690,772 B1
• Filed: 06/30/2000
• Issued: 02/10/2004
• Est. Priority Date: 02/07/2000
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Verizon Services Corporation (Verizon Communications Inc.)
• Inventors: Reformato, John, Pershan, Barry Paul, McAlily, Scott Alexander, Lubchenko, Eugene, Diede, William F., Bechtel, Kay L., Naik, Jayant M.
• Primary Examiner(s): Weaver, Scott L.
• Application Number:
  US09/608,884
• Time in Patent Office: 1,320 Days
• Field of Search: 379/67.1, 379/88.01, 379/88.02, 379/88.03, 379/88.11, 379/88.19, 379/88.2, 379/201.01, 379/207.02, 379/207.15, 379/355.06, 704/270, 704/275
• US Class Current: 379/88.03
• CPC Class Codes: H04M 3/42204 Arrangements at the exchange for service or number selection by voice at the terminal H04M1/27
• CPC Class Codes: H04M 3/4931 Directory assistance systems
• CPC Class Codes: H04Q 3/0029 Provisions for intelligent networking

Abstract

Methods and apparatus for implementing communication services such as voice dialing services are described. In one Centrex based voice dialing embodiment, voice dialing service subscribers are given access to personal voice dialing records including calling entries via the Internet as well as via telephone connections. Each calling entry normally includes the name and, optionally nickname, of a party to be called. It also includes one or more telephone numbers associated with each name. Different telephone number identifies, e.g. locations, can be associated with different names. A user can create or update entries in a voice dialing directory using text conveyed over the Internet or speech supplied via a telephone connection. In order to facilitate updating and maintenance of voice dialing directories over the Internet speaker independent (SI) speech recognition models are used. When a calling entry is created via the Internet the text of the name is processed to generate a corresponding speech recognition model there from. When an entry is created via speech obtained over the telephone, a speech recognition model is generated from the speech and a text name is generated is generated using speech to text technology. To avoid having to hang-up and initiate a new voice dialing call the outcome of a voice dialing call is monitored and the subscriber is provided the opportunity to initiate another call using voice dialing if the first call did not complete successfully e.g., goes unanswered.

Claims

1. A voice dialing method, comprising:

generating a first speech recognition model from text corresponding to a first name, the first speech recognition model being a speaker independent speech recognition model;

storing the first speaker independent speech recognition model and a first telephone number associated with the first name in a voice dialing record;

generating a second speech recognition model corresponding to a second name, the second speech recognition model being generated from speech including the second name, the step of generating a second speech recognition model including;

i) transmitting said speech, including the second name, over a telephone network to a speaker independent speech recognition model training circuit;

ii) operating the speaker independent speech recognition model training circuit to generate the second speech recognition model; and

iii) storing the second speech recognition model in said voice dialing record.

2. The method of claim 1, further comprising:

storing a second telephone number associated with the second name in said voice dialing record.

3. The method of claim 2, further comprising:

receiving speech from a caller;

using the first and second speech recognition models to perform a speech recognition operation on the received speech; and

when one of the first and second names is recognized in the received speech, placing a call to the telephone number associated with the recognized one of the first and second names.

4. The method of claim 3, further comprising:

storing information in the voice dialing record indicating that the second speech recognition model was generated from speech.

5. The method of claim 4, wherein the second speech recognition model is a speaker independent speech recognition model.

6. The method of claim 3, wherein the second speech recognition model is a speaker dependent speech recognition model.

7. The method of claim 1, further comprising:

storing information in the voice dialing record indicating that the second speech recognition model was generated from speech.

8. The method of claim 7, wherein the second speech recognition model is a speaker independent speech recognition model.

9. The method of claim 1, further comprising the step of:

operating a speech to text conversion device to generate a text version of said second name from said speech including the second name; and

storing the text version of the second name in the calling record.

10. The method of claim 9, further comprising the step of:

transmitting the text version of the second name to a subscriber'"'"'s computer via the Internet.

11. The method of claim in 10, further comprising the step of:

receiving, via the Internet, a modified text version of the second name; and

storing the modified text version of the second name in the voice dialing record.

12. The method 11, further comprising:

operating an intelligent peripheral device coupled to a telephone switch to receive additional speech from a caller;

using the first and second speech recognition models to perform a speech recognition operation on the received additional speech; and

when one of the first and second names is recognized in the received speech, placing supplying the telephone switch with the telephone number associated with the recognized one of the first and second names.

13. A method of providing a voice dialing service, the method comprising the steps of:

receiving speech from a voice dialing service subscriber including a first name;

processing the received speech to generate a first speech recognition model corresponding to the first name;

performing a speech to text operation on the received speech to generate a text version of the first name;

storing the generated speech recognition model and the text version of the first name in a voice dialing record;

presenting the text version of the first name to a voice dialing service subscriber;

receiving a modified version of the first name; and

storing the modified version of the first name in the voice dialing record.

14. The method of claim 13, wherein the step of presenting the text version of the first name includes the step of;

transmitting the text version of the first name over the Internet to a computer system; and

operating the computer system to display the text version of the first name.

15. The method of claim 14, further comprising:

receiving from the computer system telephone numbers and text versions of additional names transmitted over the Internet, at least one telephone number being associated with the text version of each additional name.

16. The method of claim 15, further comprising:

generating from the text version of each received additional name, a speaker independent speech recognition model;

storing in the voice dialing record, the text of each additional name, the at least one received telephone number associated with each additional name, and the generated speaker independent speech recognition models.

17. The method of claim 16, further comprising the step of:

storing in the voice dialing record a first model type identifier associated with the first speech recognition model, the model type identifier indicating that the first speech recognition model was generated from speech.

18. The method of claim 17, further comprising the step of:

storing in the voice dialing record a second model type identifier associated with the speaker independent speech recognition models generated from the text of each additional name, the second model type identifier indicating that the speaker independent speech recognition models with which it is associated were generated from text.

19. The method of claim 18, further comprising;

receiving speech from a caller;

using the speech recognition models stored in said voice dialing record to perform a speech recognition operation on the received speech; and

when a name is recognized in said speech using the stored speech recognition models, placing a call to the telephone number associated with the recognized one of the names.

20. A voice dialing system, comprising:

a telephone network;

an intelligent peripheral device coupled to the telephone switch and to a data network for receiving audio signals from the telephone switch and for receiving data including text via the data network, the intelligent peripheral device including;

first means for generating speech recognition models of names from text versions of names received via the data network; and

second means for generating speech recognition models of names from speech received from the telephone switch.

21. The voice dialing system of claim 20, further comprising:

means for performing a speech recognition operation on speech received from a caller using at least one speech recognition model generated by one of said first and second means for generating speech recognition models to identify a name in the received speech; and

means for supplying a telephone number corresponding to a recognized name to the telephone switch.

22. The voice dialing system of claim 21, further comprising:

a data storage device for storing speech recognition models generated by said first and second means for generating speech recognition models and for storing at least one telephone number associated with each speech recognition model.

23. The voice dialing system of claim 21, wherein the data network to which the intelligent peripheral device is coupled is the Internet.

24. The voice dialing system of claim 23, wherein the first means for generating speech recognition models includes a text to phoneme converter.

25. The voice dialing system of claim 20, further comprising:

a speech to text device for generating text versions of names from speech which is processed by the second means for generating speech recognition models to generate a speech recognition model there from.

26. A voice dialing system, comprising:

a subscriber record including;

a first set of speech recognition models generated from text versions of names;

a second set of speech recognition models generated from spoken versions of names provided by said subscriber, wherein the first and second sets of speech recognition models include speaker independent speech recognition models;

means for performing a speech recognition operation on speech using the first and second sets of speech recognition models; and

means for supplying a telephone number associated with a recognized name to a telephone switch;

wherein said subscriber record further includes;

i) text versions of names generated from speech by a speech to text operation, one of said text versions of names being associated with each of the models in the second set of speech recognition models; and

ii) a model type identifier associated with each one of the models in the first and second sets of models, the model type identifier associated with a speech recognition model indicating whether the model was generated from speech or from text.

27. A voice dialing system, comprising:

a subscriber record including;

a set of first speech recognition models generated from text versions of names;

a second set of speech recognition models generated from spoken versions of names provided by said subscriber;

means for performing a speech recognition operation on speech using the first and second sets of speech recognition models;

means for supplying a telephone number associated with a recognized name to a telephone switch; and

an integrated service control point for storing call processing records, said call processing records including information identifying voice dialing circuitry in which said subscriber record is stored.

28. The voice dialing system of claim 27, wherein the integrated service control point includes information used to provide a plurality of Centrex services, one of said Centrex services being a voice dialing service.