Speaker identification method, speaker identification device, and speaker identification system

• US 9,710,219 B2
• Filed: 06/04/2014
• Issued: 07/18/2017
• Est. Priority Date: 06/10/2013
• Status: Active Grant

• Current Assignee: Panasonic Intellectual Property Corporation of America (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Intellectual Property Corporation of America (Panasonic Holdings Corporation)
• Inventors: Matsuoka, Tomomi
• Primary Examiner(s): ALBERTALLI, BRIAN LOUIS
• Application Number:
  US14/420,749
• Publication Number:
  US 20150205568A1
• Time in Patent Office: 1,140 Days
• CPC Class Codes: G06F 3/16 Sound input; Sound output speech processing G10L
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/24 Speech recognition using non-acoustical features
• CPC Class Codes: G10L 15/25 using position of the lips, movement of the lips or face analysis
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces

Abstract

The present disclosure is a speaker identification method in a speaker identification system. The system stores registered voice signals and speaker images, the registered voice signals being respectively generated based on voices of speakers, the speaker images being respectively associated with the registered voice signals and respectively representing the speakers. The method includes: acquiring voice of a speaker positioned around a display; generating a speaker voice signal from the voice of the speaker; identifying a registered voice signal corresponding to the speaker voice signal, from the stored registered voice signals; and displaying the speaker image, which is associated with the identified registered voice signal, on the display, at least while the voice of the speaker which forms a basis of generation of the speaker voice signal is being acquired.

Claims

1. A speaker identification method in a speaker identification system which identifies voice of a speaker positioned around a display to display a result of the identification on the display,

the speaker identification system including a database which stores registered voice signals and speaker images, the registered voice signals being respectively generated based on voices of speakers, the speaker images being respectively associated with the registered voice signals and respectively representing the speakers, the method comprising:

acquiring voice of a speaker positioned around the display;

generating a speaker voice signal from the acquired voice of the speaker;

identifying a registered voice signal corresponding to the generated speaker voice signal, from the registered voice signals stored in the database; and

displaying the speaker image, which is stored in the database and is associated with the identified registered voice signal, on the display, at least while the voice of the speaker which forms a basis of generation of the speaker voice signal is being acquired,

when a correction instruction from a speaker in relation to the speaker image is received, newly acquiring voice of the speaker who has instructed the correction;

newly generating a speaker voice signal from the newly acquired voice of the speaker; and

overwriting the registered voice signal, which is stored in the database and is associated with the speaker image for which the correction instruction has been made, with the newly generated speaker voice signal, wherein

the speaker identification system includes a remote controller which has buttons to be pressed down, each of the buttons being associated previously with each of the speaker images, and

a speaker whose speaker image has been erroneously displayed on the display performs the correction instruction by speaking while pressing down the button associated with the speaker image representing the speaker whose speaker image has been erroneously displayed on the display.

2. The speaker identification method according to claim 1, wherein

the speaker image is displayed in a first display format while the voice of the speaker is being acquired, and is displayed in a second display format while the voice of the speaker is not being acquired after the voice of the speaker has been acquired.

3. The speaker identification method according to claim 1, further comprising:

detecting spoken contents of the speaker from the generated speaker voice signal; and

displaying the detected spoken contents near the speaker image, while the voice of the speaker which forms a basis of generation of the speaker voice signal is being acquired.

4. The speaker identification method according to claim 1, wherein the speaker image being displayed is erased from the display, when a prescribed time period has elapsed from the time at which the voice of the speaker which forms a basis of generation of the speaker voice signal ceases to be acquired.

5. The speaker identification method according to claim 1, wherein, of the first speaker image and the second speaker image, the speaker image which has been registered later in the database is displayed on the display in a different mode from the speaker image which has been registered earlier in the database.

6. The speaker identification method according to claim 1, wherein the correction instruction from the speaker is received in respect of the speaker image which is being displayed on the display and is not received in respect of the speaker image which is not being displayed on the display.

7. The speaker identification method according to claim 1, further comprising:

judging an attribute of the speaker from the generated speaker voice signal,

creating the speaker image based on the judged attribute of the speaker, and

storing in the database, in association with each other;

the generated speaker voice signal, the judged attribute of the speaker, and the created speaker image, wherein the generated speaker voice signal is stored in the database as the registered voice signal.

8. A speaker identification device, comprising:

a display;

a voice acquisition portion which acquires voice of a speaker positioned around the display;

a voice processor which generates a speaker voice signal from the acquired voice of the speaker;

a database which stores registered voice signals and speaker images, the registered voice signals being respectively generated based on voices of speakers, the speaker images being respectively associated with the registered voice signals and respectively representing the speakers;

an identification processor which identifies a registered voice signal corresponding to the generated speaker voice signal, from the registered voice signals stored in the database; and

a display controller which displays the speaker images, which are stored in the database and are associated with the identified registered voice signals, respectively, on the display, at least while the voice acquisition portion is acquiring each of the voices of the speakers which form a basis of generation of the speaker voice signal; and

a correction controller, wherein

the speaker identification system includes a remote controller which has buttons, each of the buttons being associated previously with each of the speaker images,

when a correction instruction from a speaker in relation to the speaker image is received, the voice acquisition portion newly acquires voice of the speaker who has instructed the correction,

the voice processor newly generates a speaker voice signal from the newly acquired voice of the speaker,

the correction controller overwrites the registered voice signal, which is stored in the database and is associated with the speaker image for which the correction instruction has been made, with the newly generated speaker voice signal, and

a speaker whose speaker image has been erroneously displayed on the display performs the correction instruction by speaking while pressing down the button associated with the speaker image representing the speaker whose speaker image has been erroneously displayed on the display.