Voice controlled messaging system and processing method

• US 6,088,428 A
• Filed: 10/22/1997
• Issued: 07/11/2000
• Est. Priority Date: 12/31/1991
• Status: Expired due to Term

• Current Assignee: Digital Sound Corp.
• Original Assignee: Digital Sound Corp.
• Inventors: Brahm, David, Trandal, David, Zhang, Xiao-Ling, Bladon, R. Anthony, Vanderford, Jan
• Primary Examiner(s): Tsang, Fan S.
• Application Number:
  US08/955,578
• Time in Patent Office: 993 Days
• Field of Search: 379/88.02, 379/88.03, 379/88.04, 379/84, 379/213, 379/214, 379/216, 379/74, 379/76, 379/245, 379/246, 379/142, 379/189, 379/196, 379/197, 379/198, 704/246, 704/247, 704/251, 704/255, 704/256
• US Class Current: 379/88.02
• CPC Class Codes: H04M 2201/40 using speech recognition
• CPC Class Codes: H04M 2203/2016 Call initiation by network rather than by subscriber
• CPC Class Codes: H04M 2242/22 Automatic class or number identification arrangements
• CPC Class Codes: H04M 3/382 using authorisation codes or passwords
• CPC Class Codes: H04M 3/42059 Making use of the calling party identifier
• CPC Class Codes: H04M 3/42068 where the identifier is used to access a profile
• CPC Class Codes: H04M 3/42102 Making use of the called party identifier
• CPC Class Codes: H04M 3/533 Voice mail systems
• CPC Class Codes: H04M 3/53383 Message registering commands or announcements; Greetings
• CPC Class Codes: H04Q 3/72 Finding out and indicating number of calling subscriber

Abstract

A voice controlled voice-messaging systems permits manipulation of voice messages over the telephone without the need for keypressed commands. Manipulation includes playback of messages, deletion of messages, saving messages, and sending a new message to a recipient from a user-defined list. Access to the messaging system is by spoken password. Voice input also allows the following functions: enrollment of a spoken password; granting or refusing user access to the system based on verification of the user speaking a password; recognition of spoken utterances of "yes" and "no" in response to prompts by the VMS; recognition of "stop" to interrupt and stop a voice message during playback; enrollment of a spoken personal directory of names of potential message recipients; and recognition of names in the spoken personal directory. The invention uses speaker verification, speaker-independent recognition (SIR) of "yes/no", speaker-independent recognition of "stop" during message playback, and speaker-dependent recognition (SDR) of a limited vocabulary of short phrases such as a list of message recipients'"'"' names.

Claims

1. A voice controlled digital voice messaging system comprising:

at least one subscriber telephone line coupled to a line interface controller means for digitizing voice signals received on the telephone line;

a mass storage device having a plurality of digitized voice prompts recorded thereon;

signal processing means coupled to a memory and to the line interface controller means and to the mass storage device for controlling storage, retrieval, playback, and sending of voice messages, comprising

means for selecting verification data responsive to an identification of the said at least one subscriber telephone line;

means for controlling access of a subscriber to the system upon verification of a spoken password of the subscriber using said selected verification data;

means for control of playback of voice messages stored on the mass storage device, comprising means for interrupting playback when the subscriber speaks a voice command during playback;

means for control of storage, retrieval, playback, and sending of voice messages in response to affirmative or negative voice commands given by the subscriber; and

means for directing messages to one of a plurality of intended message recipients, the intended recipient being identified by a spoken utterance by the subscriber of the name of the intended recipient.

2. The system of claim 1 wherein the means for controlling access to the system comprises means for speaker dependent recognition of a stored subscriber password and means for granting access to the system when the stored subscriber password is recognized.

3. The system of claim 1 wherein the means for control of playback comprises means for speaker dependent recognition of an interrupt command during playback, and means for interrupting playback of a message in response to speaker dependent recognition of the interrupt command.

4. The system of claim 1 wherein the means for control of storage comprises means for speaker independent recognition of YES or NO commands, and means for interrupting saving, erasing, replaying, and sending a message in response to speaker independent recognition of the YES or NO commands.

5. The system of claims 2, 3, or 4 wherein each means for recognition includes means for rejection of out-of-vocabulary utterances by the subscriber.

6. The system of claim 1 wherein the means for controlling access to the system further comprises means for enrollment of a speaker dependent password template, the template comprising a digital representation of features of a spoken password signal provided by the subscriber.

7. The system of claim 6 wherein the means for enrollment further comprises means for receiving a first plurality of spoken password utterances from the subscriber, means for creating the template based on a second plurality of the utterances comprising a first subset of the first plurality, and means for testing the template by comparing a third plurality of the utterances, comprising a second subset of the first plurality, to the template.

8. The system of claim 1, wherein the means for control of playback of voice messages stored on the mass storage device, comprises means for enrolling the word STOP spoken by the subscriber and means for recognition of the utterance STOP by the subscriber.

9. The system of claim 8 wherein the means for enrolling the word STOP further comprises means for receiving a first plurality of input spoken utterances from the subscriber, means for creating a template based on a second plurality of the utterances comprising a first subset of the first plurality, and means for testing the template by comparing a third plurality of the utterances, comprising a second subset of the first plurality, to the template.

10. The system of claim 1 wherein the means for control of storage, retrieval, playback and sending of voice messages further comprises means for speaker independent recognition of utterances of YES and NO.

11. The system of claim 1 wherein the means for control of storage, retrieval, playback and sending of voice messages further comprises:

means for prompting the subscriber to select a play message mode for playing messages stored on the mass storage device;

means for recognition of affirmative and negative utterances by the subscriber; and

means responsive to an affirmative utterance for playing back a stored message.

12. The system of claim 1 wherein the means for control of storage, retrieval, playback and sending of voice messages further comprises:

means for prompting the subscriber to select a save message mode for saving messages recently stored on the mass storage device;

means for recognition of affirmative and negative utterances by the subscriber; and

means responsive to an affirmative utterance for saving a recently received stored message on the mass storage device.

13. The system of claim 1 wherein the means for control of storage, retrieval, playback and sending of voice messages further comprises:

means for prompting the subscriber to select an erase message mode for erasing messages recently stored on the mass storage device;

means for recognition of affirmative and negative utterances by the subscriber; and

means responsive to an affirmative utterance for erasing a recently received message from the mass storage device.

14. The system of claim 1 wherein the means for control of storage, retrieval, playback and sending of voice messages further comprises:

means for prompting the subscriber to select a review saved message mode for replaying messages previously stored on the mass storage device by the subscriber;

means for recognition of affirmative and negative utterances by the subscriber; and

means responsive to an affirmative utterance for replaying a message stored on the mass storage device.

15. The system of claim 1 wherein the means for control of storage, retrieval, playback, and sending of voice messages further comprises:

means for prompting the subscriber to select an erase saved message mode for erasing a message stored on the mass storage device by the subscriber in the save message mode;

means for recognition of affirmative and negative utterances by the subscriber; and

means responsive to an affirmative utterance for erasing a saved message from the mass storage device.

16. The system of claim 1 wherein the means for control of storage, retrieval, playback, and sending of voice messages further comprises:

means for prompting the subscriber to select a send message mode for sending messages from the subscriber to one of a plurality of intended message recipients;

means, responsive to the means for prompting the subscriber, for sending messages from the subscriber to one of the plurality of intended message recipients, comprising;

means for recognition of the name of an intended message recipient; and

means for recording a digitized voice message to the intended recipient on the mass storage device.

17. The system of claim 16, wherein the means for sending messages from the subscriber further comprises means for selectively replaying the voice message upon speaker independent recognition of a YES utterance from the subscriber.

18. The system of claim 16 wherein the means for sending messages further comprises means for enrolling a plurality of speaker dependent templates representing digitized utterances of a plurality of names of intended message recipients.

19. The system of claim 9 wherein the template is created using a hidden Markov model having at least fourteen states and having cepstral coefficients of at least twentieth order.

20. The system of claim 19 wherein the means for testing the template comprises means for measuring spectral distance using smoothed Laplacian group delay spectrum.

21. The system of claim 19 wherein the means for testing the template comprises means for weighting the cepstral coefficients using smoothed group delay.

22. The system of any of claims 8, 10, 11, 12, 13, 14, 15, 16, or 17 wherein the means for recognition comprises means for creating a digital template and means for comparing the digital template to the spoken password of the subscriber, the template representing an earlier measured utterance of the password by the subscriber, using a hidden Markoff model, having at least 14 states and having cepstral coefficients of at least 20th order.

23. The system of claim 22 wherein the means for comparing the template comprises means for measuring spectral distance using smoothed Laplacian group delay spectrum.

24. The system of claim 23 wherein the means for comparing the template comprises means for weighting the cepstral coefficients using smoothed group delay.

49. The system of claim 18 wherein the speaker dependent templates are created using a hidden Markov model having at least fourteen states and having cepstral coefficients of at least twentieth order.

25. In a data processing system having at least one subscriber telephone line coupled to a line interface controller for digitizing voice signals received on the telephone line, a mass storage device having a plurality of digitized voice prompts recorded thereon, and signal processing means coupled to a memory and to the line interface controller and to the mass storage device;

a method for controlling storage, retrieval, playback, and sending of voice messages, comprising the steps of;

identifying the at least one subscriber telephone line;

selecting verification data responsive to said identification;

providing access to the system in response to verification of a spoken password uttered by a subscriber, said verification using the selected verification data;

selectively playing back voice messages stored on the mass storage device, comprising the substep of interrupting playback in response to an utterance by the subscriber of a voice command during playback;

controlling storage, retrieval, playback and sending of voice messages in response to utterances of affirmative or negative voice commands by the subscriber; and

selectively sending messages to one of a plurality of intended message recipients, the intended recipient being identified by a spoken utterance by the subscriber of the name of the intended recipient.

26. The method of claim 25 wherein the step of providing access to the system further comprises the step of recording a speaker dependent password template, the template comprising a digital representation of features of a spoken password signal provided by the subscriber.

27. The method of claim 26, wherein the step of providing access to the system further comprises the steps of comparing the template to a second template formed from an utterance by the subscriber, determining whether the templates match, and in response to a template match, granting access to the system.

28. The method of claim 25 wherein the affirmative utterance is YES and the negative utterance is NO.

29. The method of claim 25 wherein the step of providing access to the system further comprises the steps of:

receiving a first plurality of input spoken password utterances from the subscriber,

creating a template based on a second plurality of the utterances comprising a first subset of the first plurality,

testing the template by comparing a third plurality of the utterances, comprising a second subset of the first plurality, to the template,

and in response to the testing step, rejecting the template if the third plurality of utterances do not match the template.

30. The method of claim 25, wherein the step of selectively playing back voice messages stored on the mass storage device, includes the substep of enrolling an interrupt command word spoken by the subscriber.

31. The method of claim 30, wherein the interrupt command word is STOP.

32. The method of claim 25, wherein the step of controlling storage, retrieval, playback, and sending includes the substep of enrolling an interrupt command word spoken by the subscriber.

33. The method of claim 32, wherein the interrupt command word is STOP.

34. The method of claims 30 or 32 wherein the substep of enrolling the interrupt command word further comprises the steps of:

receiving a first plurality of input spoken password utterances from the subscriber,

creating a template based on a second plurality of the utterances comprising a first subset of the first plurality,

testing the template by comparing a third plurality of the utterances, comprising a second subset of the first plurality, to the template,

and in response to the testing step, rejecting the template if the third plurality of utterances do not match the template.

35. The method of claim 25 wherein the step of controlling storage, retrieval, playback, and sending of voice messages further comprises the step of speaker independent recognition of affirmative or negative command utterances.

36. The method of claim 25 wherein the step of controlling storage, retrieval, playback and sending of voice messages further comprises the steps of:

prompting the subscriber to select a play message mode for playing messages stored on the mass storage device;

recognition of affirmative or negative utterances by the subscriber; and

responding to an affirmative utterance by playing back a stored message.

37. The method of claim 36 wherein the step of recognition further comprises the step of rejecting out-of-vocabulary utterances.

38. The method of claim 25 wherein the step of controlling storage, retrieval, playback and sending of voice messages further comprises the steps of:

prompting the subscriber to select a save message mode for saving messages recently stored on the mass storage device;

recognition of affirmative or negative command utterances by the subscriber; and

responding to an affirmative utterance for saving a recently received stored message on the mass storage device.

39. The method of claim 25 wherein the step of controlling storage, retrieval, playback, and sending of voice messages further comprises the steps of:

prompting the subscriber to select an erase message mode for erasing messages recently stored on the mass storage device;

recognition of affirmative or negative utterances by the subscriber; and

responding to an affirmative utterance by erasing a recently received message from the mass storage device.

40. The method of claim 25 wherein the step of controlling storage, retrieval, playback, and sending of voice messages further comprises:

prompting the subscriber to select a review saved message mode for replaying messages previously stored on the mass storage device by the subscriber;

recognition of affirmative or negative utterances by the subscriber; and

responding to an affirmative utterance for replaying a message stored on the mass storage device.

41. The method of claim 25 wherein the step of controlling storage, retrieval, playback and sending of voice messages further comprises the steps of:

prompting the subscriber to select an erase saved message mode for erasing messages stored on the mass storage device by the subscriber in the save message mode;

recognition of affirmative or negative utterances by the subscriber; and

responding to an affirmative utterance by erasing a saved message from the mass storage device.

42. The method of claim 25 wherein the step of controlling storage, retrieval, playback and sending of voice messages further comprises the steps of:

prompting the subscriber to select a send message mode for sending messages from the subscriber to one of a plurality of intended message recipients;

recognition of affirmative or negative utterances by the subscriber;

responding to an affirmative utterance for sending messages from the subscriber to one of the plurality of intended message recipients, comprising the steps of;

recognition of the name of an intended message recipient spoken by the subscriber; and

recording a digitized voice message of the subscriber to the intended recipient on the mass storage device.

43. The method of claim 42, wherein the step of responding to an affirmative utterance further comprises the substeps of prompting the subscriber whether to replay the message, and selectively replaying the message upon recognition of an affirmative utterance from the subscriber.

44. The method of claim 25 further comprising the step of enrolling a plurality of speaker dependent templates representing digitized utterances of a plurality of names of intended message recipients prior to the step of sending messages.

45. The method of claim 44 wherein the speaker dependent templates are created using a hidden Markov model having at least fourteen states and having cepstral coefficients of at least twentieth order.

46. The method of claim 45 further comprising testing the speaker dependent templates by measuring spectral distance using smoothed group delay spectrum.

47. The method of claim 45 further comprising testing the speaker dependent templates by weighting the cepstral coefficients using smoothed group delay.

48. A voice controlled digital voice messaging system comprising:

at least one subscriber telephone line coupled to a line interface controller means for digitizing voice signals received on the telephone line;

a mass storage device having a plurality of digitized voice prompts recorded thereon;

signal processing means coupled to a memory and to the line interface controller means and to the mass storage device for controlling storage, retrieval, playback, and sending of voice messages, comprising

means for selecting verification data responsive to an identification of the said at least one subscriber telephone line;

means for controlling access of a subscriber to the system upon verification of a spoken password of the subscriber using said selected verification data;

means for control of playback of voice messages stored on the mass storage device, comprising means for interrupting playback when the subscriber speaks a voice command during playback; and

means for directing messages to one of a plurality of intended message recipients, the intended recipient being identified by a spoken utterance by the subscriber of the name of the intended recipient.