Visualization interface of continuous waveform multi-speaker identification

• US 8,347,247 B2
• Filed: 10/17/2008
• Issued: 01/01/2013
• Est. Priority Date: 10/17/2008
• Status: Active Grant

• Current Assignee: Globalfoundries US Incorporated (GlobalFoundries, Inc.)
• Original Assignee: International Business Machines Corporation
• Inventors: Harrington, Nathan J.
• Primary Examiner(s): Chawan, Vijay B
• Application Number:
  US12/253,472
• Publication Number:
  US 20100100376A1
• Time in Patent Office: 1,537 Days
• Field of Search: 716/113, 716/108, 716/136, 716/109, 716/115, 701/102, 701/101, 379/202.01, 379/88.02, 703/2, 703/13, 703/14, 340/440.2, 702/67, 702/71, 380/263, 380/46, 704/273, 704/270, 704/231, 704/235, 706/14, 707/661, 707/707, 707/827
• US Class Current: 716/113
• CPC Class Codes: G06F 16/683 using metadata automatically derived from the content
• CPC Class Codes: G06F 18/2135 based on approximation criteria, e.g. principal component analysis
• CPC Class Codes: G06F 21/32 using biometric data, e.g. fingerprints, iris scans or voiceprints
• CPC Class Codes: G10L 13/08 Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

A method implemented in a computer infrastructure having computer executable code having programming instructions tangibly embodied on a computer readable storage medium. The programming instructions are operable to receive a current waveform of a communication between a plurality of participants. Additionally, the programming instructions are operable to create a voiceprint from the current waveform if the current waveform is of a human voice. Furthermore, the programming instructions are operable to determine one of whether a match exists between the voiceprint and one library waveform of one or more library waveforms, whether a correlation exists between the voiceprint and a number of library waveforms of the one or more library waveforms and whether the voiceprint is unique. Additionally, the programming instructions are operable to transcribe the current waveform into text and provide a match indication display (MID) indicating an association between the current waveform and the one or more library waveforms based on the determining.

Claims

1. A method implemented in a computer infrastructure having computer executable code tangibly embodied on a computer readable storage medium having programming instructions operable to:

receive a current waveform of a communication between a plurality of participants;

create a voiceprint from the current waveform if the current waveform is of a human voice;

determine one of;

whether a match exists between the voiceprint and one library waveform of one or more library waveforms;

whether a correlation exists between the voiceprint and a number of library waveforms of the one or more library waveforms; and

whether the voiceprint is unique;

transcribe the current waveform into text; and

provide a match indication display (MID) indicating an association between the current waveform and the one or more library waveforms based on the determining.

2. The method of claim 1, further comprising displaying the current waveform and the one or more library waveforms.

3. The method of claim 2, further comprising associatively displaying the text with its associated library waveform.

4. The method of claim 1, wherein the indicating the association comprises displaying at least one of a match percentage and an indicator arrow.

5. The method of claim 1, wherein at least one of the receiving, the creating, the determining, the transcribing and the providing occurs in real-time.

6. The method of claim 1, wherein the receiving is via a single microphone.

7. The method of claim 1, further comprising verifying whether the current waveform is of a human voice, wherein the verifying whether the received current waveform is of the human voice comprises determining whether approximately ninety percent or more of frequencies of the current waveform occur in a frequency band from approximately 200 Hz-800 Hz.

8. The method of claim 1, further comprising:

adding the voiceprint to the one or more library waveforms if the voiceprint is unique; and

averaging the voiceprint with the one library waveform if the match exists between the voiceprint and the one library waveform of the one or more library waveforms.

9. The method of claim 1, wherein the creating the voiceprint from the current waveform comprises:

dividing the current waveform into equal lengths of frequency range;

determining an average amplitude of the current waveform for each of the equal lengths of frequency range;

storing the average amplitude of the current waveform for each of the equal lengths of frequency range as the voiceprint in a storage system; and

updating the average amplitude of the current waveform for each of the equal lengths of frequency range.

10. The method of claim 9, wherein the determining whether the match or the correlation exists or whether the voiceprint is unique comprises:

determining an integer difference between the average amplitude of the current waveform and the one or more library waveforms for each of the equal lengths of frequency range;

if the integer difference for each of the equal lengths of frequency range for the one library waveform of the one or more library waveforms is less than a match threshold, determining that the match exists between the voiceprint and the one library waveform of the one or more library waveforms;

if the integer difference for at least one of the equal lengths of frequency range for each of the library waveforms is more than a unique threshold, determining that the voiceprint is unique; and

if the voiceprint does not meet the match threshold and does not meet the unique threshold, determining that the correlation exists between the voiceprint and the number of library waveforms of the one or more library waveforms based on the integer difference for each of the equal lengths of frequency range.

11. The method of claim 10, wherein a degree of the correlation between the voiceprint and the number of library waveforms is displayed as a match percentage for each of the number of library waveforms.

12. The method of claim 11, further comprising

determining a library waveform of the number of waveforms having a highest match percentage;

determining a ratio of remaining match percentages for each of the number of library waveforms to the match percentage of the library waveform having the highest match percentage; and

if the ratio for each of the number of library waveforms do not exceed a relative weight threshold, associatively displaying the current waveform with the library waveform having the highest match percentage.

13. The method of claim 12, wherein the MID further provides an unassigned transcribed text window;

and

if the ratio for any of the number of library waveforms do exceed the relative weight threshold, displaying the transcribed text of the current waveform in the unassigned transcribed text window.

14. The method of claim 1, wherein the MID further provides a temporal indication for the text.

15. The method of claim 14, wherein the temporal indication for the text comprises at least one of:

indicator arrows which change colors based on temporal proximity to current time;

indicator arrows which fade or become translucent based on temporal proximity to current time; and

sequential numbers associatively displayed with the text.

16. The method of claim 1, wherein a service provider at least one of creates, maintains, deploys and supports the computer infrastructure.

17. The method of claim 1, wherein steps are provided by a service provider on a subscription, advertising, and/or fee basis.

18. The method of claim 1, wherein the transcribing the current waveform into text occurs in real time.

19. A system, comprising:

a human voice verification (HVV) tool operable to verify whether a received current waveform is of a human voice;

a voiceprint creation (VPC) tool operable to create a voiceprint from the received current waveform if the received current waveform is of the human voice;

a matching tool operable to determine one of;

whether a match exists between the voiceprint and one library waveform of one or more library waveforms;

whether a correlation exists between the voiceprint and a number of library waveforms of the one or more library waveforms; and

whether the voiceprint is unique; and

a display tool operable to;

transcribe the received current waveform into text; and

provide a match indication display (MID) indicating an association between the received current waveform and the one or more library waveforms based on the determining of the matching tool.

20. The system of claim 19, wherein the MID is further operable to:

display the received current waveform and the one or more library waveforms; and

associatively display the text with its associated library waveform.

21. The system of claim 19, wherein the indicating the association comprises displaying at least one of a match percentage and an indicator arrow.

22. The system of claim 19, wherein at least one of the HVV tool, the VPC tool, the matching tool and the displaying tool operate in real-time.

23. The system of claim 19, wherein the MID further provides a temporal indication for the text comprising at least one of:

an indicator arrow which change colors based on temporal proximity to current time;

an indicator arrow which fade or become translucent based on temporal proximity to current time; and

sequential numbers associatively displayed with the text.

24. The system of claim 19, wherein the transcribing the current waveform into text occurs in real time.

25. A computer program product comprising a computer usable storage medium having readable program code embodied in the storage medium, the computer program product includes at least one component operable to:

receive a current waveform of a communication between a plurality of participants;

create a voiceprint from the current waveform;

determine one of;

whether a match exists between the voiceprint and one library waveform of one or more library waveforms;

whether a correlation exists between the voiceprint and a number of library waveforms of the one or more library waveforms; and

whether the voiceprint is unique;

transcribe the current waveform into text; and

provide a match indication display (MID) displaying the text and indicating an association between the current waveform and its associated text, and the one or more library waveforms based on the determining.

26. The computer program product of claim 25, wherein the transcribing the current waveform into text occurs in real time.

27. A method comprising:

providing a computer infrastructure operable to;

receive a current waveform of a communication between a plurality of participants;

verify whether the current waveform is of a human voice;

create a voiceprint from the current waveform if the current waveform is of the human voice;

determine one of;

whether a match exists between the voiceprint and one library waveform of one or more library waveforms;

whether a correlation exists between the voiceprint and a number of library waveforms of the one or more library waveforms; and

whether the voiceprint is unique;

transcribe the current waveform into text; and

provide a match indication display (MID) displaying the text and indicating an association between the current waveform and its associated text, and the one or more library waveforms based on the determining,

wherein at least one of the receiving, the verifying, the creating, the determining, the transcribing and the providing occurs in real-time.

28. The method of claim 27, wherein the computer infrastructure is operable to receive the current waveform of the communication between the plurality of participants via a single microphone.

29. The method of claim 27, wherein the transcribing the current waveform into text occurs in real time.