JPS602997A - Speaker identifying voice inputting method - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

Detailed Description of the Invention (1) Technical Field of the Invention The present invention recognizes the voice of an input speaker by storing standard patterns for multiple speakers in a voice input device for each speaker and specifying the speaker. This paper relates to a voice input method that makes it possible to specify a speaker by remote control using a device that uses the device.

(2) Conventional technology and problems Conventional voice input devices wait for the standard patterns of multiple speakers for each speaker, and then specify the ``#'' before recognition. It is often used to store things in

In this case, when standard patterns for multiple speakers are registered, a hardware switch is provided to select a speaker, or a speaker is specified from a thousand boards. A more commonly used method is to have a floppy disk containing the standard pattern for each ti-i1, and to insert it at the start of work. However, with these methods, when switching between standard patterns, it is necessary to go to the TFT of the voice input device and operate it, and the remote edge 1'
I couldn't form a ridge.

(3) Purpose of the Invention The purpose of the present invention is to provide a device for recognizing standard patterns of human speakers by storing four standard patterns of virtual speakers and specifying the speaker. In addition to specifying the speaker in a ttsw format, it also provides a voice input method for each speaker that can be controlled remotely.

(4) Structure of the Invention In order to achieve all of the above objects, the speaker identification voice input method of the present invention has a standard pattern t for each speaker, and stores the standard pattern corresponding to the speaker in the recognition device. Particularly, in a speech recognition device that can recognize the voice of an input speaker, the name of each speaker is registered in a part of the standard pattern area of each speaker, and the input speaker can input his/her name. The feature is that one's own standard pattern can be stored in the recognition device by voice input.

(5) Embodiments of the Invention In the present invention, it is basically necessary to use voice to switch the standard pattern performance ranges of six speakers, thereby enabling remote P4 operation.

In the present invention, the speaker is designated by voice, and at 5fc, a reference section in which all the standard patterns of names uttered by the valley speaker is arranged is provided in a part of the valley speaker's standard pattern area.

This was developed to facilitate immediate access when the voice of the name defined by the speaker tf'tx is input.

FIGS. 1 (L) to (c) and FIGS. 2 (1) to (2) are explanatory views of the method of registering standard patterns for each speaker according to the present invention.

FIG. 1 (α) ■ to ■ show the standard pattern area 1 of speaker A. An index area 2 is provided at the top of the area, and standard patterns of names pronounced by all speakers A, B, and C are arranged in this area. At the bottom is speaker A's q! R types of required standard patterns 3 are stored. To register this, register the part A of the index reconnaissance area 2 and the voice pattern 3 of A to speaker A in (α)■ in the same figure, and
Register part B of the index area B and part C of the index area for speaker C in (α)■ in the same figure, and generate the standard pattern shown in (α)■ in the same figure. This is the standard pattern for speaker A.

Figures ('b) and (c) are standard patterns for speakers B and C, respectively.

FIGS. 2-2 are explanatory diagrams of other registration methods.

In the same figure, ■ to ■ are for speakers A, B, and C.Same as in Figure 1 (α)■, the standard pattern of the name in each voice is placed in index area 2, and the required standard pattern for each building is placed in area 3. This is registered and mutually transferred and written in the blank column of the name using a knot (program) on the memory 10 shown in (2) in the same figure.

FIG. 6 is an explanatory diagram of the configuration of an embodiment of the present invention.

In the figure, the voice input device is controlled by a controller 11, and an fc7 isle 10. Softfare 12. Operating device 1
62 is constructed from the display section 14 and connected to the recognition device 16 via the controller 11. File 1o contains figures 1 and 2.
For each story registered by the method shown in the figure, a standard pattern area for each f is stored with an index area of the standard pattern / of the names of all speakers at the beginning.

The control unit 11 controls the standard patterns for each speaker in the seven aisles 1o when they are registered, when they are recognized, and when they are recognized. In this case, control according to a predetermined program is performed by the software 12, and manual control and monitoring by an operator are performed using the operating device 13 and the display unit 14. At the time of registration and recognition, the voice of the speaker is inputted into the microphone 15 and into the recognition device 16, where a standard pattern of Kugata's voice is created. When replacing the standard pattern of the recognition device, enter the speaker's name audibly, enter the recognition code corresponding to the name into the controller 211, and enter the recognition code from the standard patterns/for each speaker in the file 10. In response to this, the contents of the standard and customary areas of the speaker, for example, Bo, are collectively transferred to the recognition device 16 via the controller 11 and buffered to correspond to the following voice input from one speaker B. The recognition device 16 may be directly connected in the vicinity of the voice input device, or may be provided at a remote location and connected on the same line,
Registration, recognition.

Mode designation such as speaker designation can be remotely controlled.

FIG. 4 is a flowchart showing the operation of the embodiment shown in FIG. 3 during recognition.

In the same figure, the controller 11t-talk'4 designation mode is set (by the mode designation control of 16), and the speaker inputs the speaker's name t-to the recognition device 16 through the microphone/15. Here, a voice pattern of the speaker's name is created, sent to the file 10 through the entire controller 11, and the above-mentioned index is performed to check whether there is corresponding registered data. G-weaving device 16
If the voice input is too small or below, it will be recognized by comparing it with this content.

(6) As described in detail, according to the present invention, the speaker selects the content of the corresponding standard pattern area in the voice input device by vocally inputting his or her name. Since the voice input device is used for the following voice inputs, there is no need to go to the voice input device by yourself, the operation can be done in the dark, and remote control is easily possible. Therefore, it is expected that the present invention will be useful in expanding the scope of use of voice input devices.

[Brief explanation of drawings]

FIG. 1 and FIG. 2 are explanatory diagrams of the phonetic dictionary 4 registration method of the present invention,
FIG. 6 is a configuration explanatory diagram of an embodiment of the present invention, and FIG. 4 is a flowchart showing the operation during recognition of the present invention. In the figure, 1 is a voice pattern area for each speaker, 2 is an index section, and 6 is the voice pattern capital,
10 is an audio file) 11 is a controller, 12 is software, 16 is an operating device, 14 is a display section, 15 is a microphone 116 is a recognition device. Patent Applicant Fujitsu Kinsha Sub-Agent Patent Attorney 1) Zen Saka Mold Opening Figure 1 (b) (c) Figure 2 Figure 3 Figure 4

Claims (1)

[Claims] Standard patterns for multiple speakers In a speech recognition device that can recognize the voice of an input speaker by transmitting a standard pattern corresponding to each speaker to the recognition device, one of the standard pattern areas for a valley speaker is used. One in the department? ! i Record each speaker's name in voice, and when the input speaker enters his or her name on the back door, p his or her standard pattern is input to the recognition device, f
6. A voice input method for different types of voice input, which is characterized by the ability to use 6 networks.