JP2001067095A - Voice recognizing method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a voice recognizing method capable of identifying the voice of a user without being affected by ambient noise. SOLUTION: The voice of a user and dial information are previously memorized in a vocal telephone directory-memory part 19 and ambient noise, in an ambient noise-memory part 25. Also in use, the voice of the user and the ambient noise are inputted from a handset 9 and from a main-body microphone 26, respectively. A standard pattern reproducing part 22 finds differences between the voice of the user, registered previously and inputted in use, and the ambient noise, thereby reproducing net voice. Similar data in the vocal telephone directory-memory part 19 is determined and extracted by a similarity determining part 21, thereby ensuring voice recognition, and an automatic call can be made based on the registered dial information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech recognition method and apparatus using pure speech information.

[0002]

2. Description of the Related Art FIG. 7 is a block diagram of a conventional speech recognition apparatus, FIG. 8 is a flowchart of speech recognition using the speech recognition apparatus, and FIG. 5 shows a flowchart for inputting voice information to a memory unit.

The operation of speech recognition will be described with reference to FIGS.

In the conventional voice recognition, voice information is stored in a voice telephone directory / memory unit 19 in advance. When the user inputs a voice when using the device with the voice recognition device, the similarity determination unit 21 compares the voice with the voice information stored in the voice telephone directory / memory unit 19.
It is a specification that makes a judgment and obtains a similarity judgment result.

The operation will be described with reference to a flow chart. When voice recognition is started (step 301), voice input is instructed by fixed guidance (step 302), and voice input is performed (step 303). Next, the voice information stored in the voice telephone directory / memory unit 19 is read out (step 304).
The similarity determination unit 21 compares and determines with the input voice (step 305).

Next, as a method of storing voice information in the telephone directory / memory unit 19, conventionally, a flow of FIG. 9 is used. When voice registration is started (step 401), an instruction to input voice from the memory input unit is issued by fixed guidance (step 402). The user inputs voice (step 403) and stores it as voice information in the voice telephone directory / memory unit 19 (step 404).

FIG. 11 shows a configuration diagram of a conventional key telephone system with a voice recognition function. Reference numeral 1 denotes a key telephone device, which comprises a central office interface unit 2, an extension interface unit 3, a main power supply unit 5, and a system control unit 4 for controlling these. Reference numeral 6 denotes a key telephone connected to the key telephone device 1, which includes a communication circuit unit 8 for controlling external communication, a handset 9 having an input unit,
A key matrix unit 10 for key input, an LED display unit 11, an LCD display unit 12, a slave unit power supply unit 13, a voice recognition unit 20 for performing voice recognition, a main unit microphone 26 as an input unit, a speaker 27, and control thereof. It is composed of a slave unit interface unit 7. The voice recognition unit 20 includes an A / D conversion unit 17, a fixed guidance / memory unit 18, a voice telephone directory / memory unit 19, a similarity determination unit 21, and a voice recognition control unit 16 that controls these to perform voice recognition. Have been.

The basic operation of the key telephone system will be described. When the key telephone 6 is connected to the key telephone device 1,
Power is supplied to the slave unit power supply unit 13 by the main power supply unit 5 via the extension interface unit 3, and the key telephone 6 becomes usable. When the handset 9 of the key telephone 6 is off-hook and the dial is pressed, dial information is detected by the key matrix unit 10 and transmitted to the system control unit 4 via the slave unit interface unit 7 and the extension interface unit 3. . If the dial information is for requesting a local line call, the system control unit 4 connects the telephone line between the central line interface unit 2 and the extension interface unit 3, and enables a telephone call with the handset 9 via the telephone line circuit unit 8. Become. In the case of hands-free, a telephone call with the main unit microphone 26 and the speaker 27 becomes possible. Further, under the control of the system control unit 4, it is possible to display station line communication state information and the like on the LED display unit 11 and the LCD display unit 12. That is, if the user picks up the handset 9 and presses the dial, it can be used as a normal telephone.

The following is a specific example of the use of FIG. In other words, when the voice is input as "Suzuki" from the handset 9 or the main unit microphone 26 in accordance with the instruction of "Please say your name" stored in the fixed guidance memory unit 18, the voice recognition control unit 16 recognizes “Suzuki”, and extracts information about “Suzuki” from the information registered in the voice telephone directory / memory unit 19 by the similarity determination unit 21. And
Information such as "Suzuki A man" and "Suzuki B child" are associated with numbers and displayed on the LCD display unit 12 as shown in FIG. Subsequently, the user is prompted for confirmation by the fixed guidance in the fixed guidance memory unit 18 ("Who is the other party?"). When the user voice-inputs "1" or "1" from the key matrix unit 10, the voice information ("Suzuki A") in the voice telephone directory / memory unit 19 is input.
Dial information (“045”) stored with the “man”
-***-〇 × △ □ ”) to the system controller 4.
It is controlled through the central office line interface unit 2 to make a dial call. That is, when the name of the other party to be called is input by voice, registered data (the name and the telephone number) is displayed, and when the user accepts by voice or key input, an automatic call is made to the telephone number. It is.

[0010] It is not necessary to display the telephone number because of confidentiality.

[0011]

However, in the conventional telephone apparatus with a voice recognition function, voice information is stored in the voice telephone directory / memory unit 19 by voice input including the background noise at the time of registration. Of course, they are input at the same time. Even if there is no ambient noise at the time of registration, the surrounding state at the time of registration and the surrounding state at the time of use do not become the same, such as being in a noisy crowd at the time of use. Therefore, there is a problem that the voice of the user cannot be correctly recognized at the time of use.

An object of the present invention is to solve such a conventional problem, and an object of the present invention is to provide an excellent voice recognition method and apparatus which can surely recognize a user's voice.

[0013]

In order to solve the above-mentioned problem, a voice recognition method according to the present invention stores ambient noise in advance to obtain a difference between an input voice and the ambient noise, and stores the difference based on the difference. This is a method for searching and extracting similar voices from the voice information stored in the.

According to this method, a pure voice from which the ambient noise has been removed is extracted from the voice input including the ambient noise, and can be compared and determined with the voice stored in the storage means in advance, without being affected by the ambient noise. And a voice recognition function can be improved.

The voice recognition method of the present invention is a method in which a voice from which ambient noise has been subtracted in advance is stored in a storage means as voice information.

According to this method, pure speech can be used as basic data.

Further, the voice recognition device of the present invention has a voice input means for inputting ambient noise and a user's voice, a first storage means for storing only the ambient noise, and a voice input means for inputting from the voice input means. Voice extracting means for calculating a difference between the voice of the user and the ambient noise stored in the first storage means;
A second storage unit in which audio information is stored in advance, and a comparison unit that compares / determines the audio extracted by the audio extraction unit with the audio information stored in the second storage unit. .

According to this configuration, the voice recognition method described in claim 1 can be realized, and security can be improved by applying a telephone device, an unlocking of a front door, and the like.

Further, the voice extracting means includes a third storage means for storing the voice input from the voice input means, and obtains a difference between the contents of the first and third storage means. Things.

Thus, the input voice can be directly stored and can be used for updating the stored data as needed.

Further, as the audio information of the second storage means,
It is characterized by using the voice obtained from the voice extracting means.

This makes it possible to provide a speech recognition device for comparing pure speech information stored in advance with pure input speech.

[0023]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to FIGS. The same components as those in the conventional example are denoted by the same reference numerals, and description thereof is omitted.

FIG. 1 is a block diagram of a voice dial transmission operation in the voice recognition device of the first embodiment, FIG. 2 is a flowchart of voice recognition using this voice recognition device, and FIG. FIG. 4 shows a block diagram of a voice dial registration operation in the voice recognition device of the first embodiment,
The operation will be described with reference to the flowchart of FIG. FIG. 5 is a configuration diagram of a key telephone system with a voice recognition function according to the first embodiment. FIG. 6 is a configuration diagram of a single telephone with a voice recognition function according to the first embodiment.
It is a voice telephone directory / memory unit that is a second storage unit. A standard pattern reproducing unit 22 and an ambient noise / memory unit 25 serving as a first storage unit are further provided in the voice recognition unit 20 of the conventional example.
Is added.

The operation of the speech recognition according to the first embodiment will be described with reference to FIGS.

First, only the ambient noise is input and stored in the ambient noise / memory section 25 as the first storage means. Next, in the standard pattern reproducing unit 22, a difference between the input user's voice and the ambient noise data stored in the ambient noise / memory unit 25 is obtained, and the difference is stored in the second storage unit in advance. The similarity determination unit 21 compares and determines the audio information that has been used, and obtains a similarity determination result.

The operation will be described with reference to a flow chart. When voice recognition (voice dialing operation) is started (step 101), silence is first instructed (step 102), ambient noise data is input (step 103), and this ambient noise data is first stored. It is stored in the ambient noise / memory unit 25 (step 104). Subsequently, a voice input of the user is requested (step 105), and when the voice is input while the ambient noise is included from the user (step 106), the two differences are obtained in the standard pattern reproducing unit 22 (step 106). 107). Then, the voice information stored in advance in the voice telephone directory / memory section 19 as the second storage means is called (step 108), and the voice information and the voice reproduced by the standard pattern reproducing section 22 are called out. The similarity determination unit 21 repeatedly performs similarity determination (step 109) and extracts the same data (step 1).
10). Then, it is displayed on the LCD display section 12 (step 1).
11) Then, the user is caused to specify the destination by the fixed guidance (step 112). Dialing is performed based on the contents specified by the user (step 113). The display data in step 112 is the same as in the conventional example as shown in FIG.

As described above, according to the first embodiment of the present invention, only the ambient noise is inputted and stored in the ambient noise / memory section 25 as the first storage means. The difference between the ambient noise data in the ambient noise / memory unit 25 and the voice input by the user is obtained, and this difference is compared with the voice telephone directory data.
By comparing the sound information with the sound information in the memory unit 19, the sound can be reliably recognized without being affected by the ambient noise.

Further, in the first embodiment of the present invention,
Although a key telephone system for connecting a key telephone as shown in FIG. 5 is described as an example, the present invention is also applicable to a single telephone with a voice recognition function shown in FIG. In FIG. 6, reference numeral 23 denotes a control unit for performing various controls, and reference numeral 25 denotes an ambient noise / memory unit as ambient noise storage means. In addition, the present invention can be applied to items related to security management such as unlocking of a front door.

It is not necessary to display the telephone number for security. If there is no corresponding data, voice recognition cannot be performed, but it can be used as a normal call mode.

Other details not related to the present invention will not be described in detail.

The operation of voice registration used for voice recognition according to the first embodiment will be described with reference to FIGS.

First, only the ambient noise is inputted and stored in the ambient noise / memory section 25 as the first storage means. Next, in the standard pattern reproducing unit 22, the difference between the input user's voice and the ambient noise data stored in the ambient noise / memory unit 25 is obtained, and the voice telephone directory / memory as the second storage unit is obtained. The specifications are stored in the unit 19.

The operation will be described with reference to a flow chart. When voice registration (voice dial registration operation) is started (step 201), first, the fixed guidance memory unit 18 instructs the user to remain silent for a while (step 2).
02). While the user is not speaking, ambient noise (ambient noise data) is input from the handset 9 or the microphone 26 (step 203), and the ambient noise is stored in the ambient noise / memory unit 25 as the first storage unit. It is stored as noise data (step 204). Subsequently, a voice input of the user is requested (step 205), and voice input is performed while including the ambient noise from the user (step 206).
The standard pattern reproducing unit 22 calculates the difference between the two, and reproduces the pure voice of the user (step 207). Then, the reproduced pure voice of the user is stored as voice information in the voice telephone directory / memory unit 19 as the second storage means (step 208). The user is required to enter the telephone number in the fixed guidance (step 20).
9) When input from the user (step 210),
The telephone number is stored in the voice telephone directory / memory unit 19 in correspondence with the voice information (step 211). At this time,
Of course, the ambient noise data and the voice information of the user are:
It is associated and stored.

According to a second embodiment of the present invention, as set forth in claim 4, there is provided a third storage means for storing a voice input while including ambient noise. The voice input from the user is stored as it is (including the ambient noise), and the user's voice input including the input ambient noise (stored in the third storage means) and the ambient noise (the (Stored in the first storage means) to determine the similarity and determine the similarity. In this case, in the flowchart of FIG.
Step 106 → step 121 → step 122 → step 108. In the flowchart of FIG. 4, step 206 is not performed and step 206 → step 221 → step 222 → step 208.

[0036]

As described above, according to the present invention, only the ambient noise is input and stored in the first storage means, and the difference between the input voice and the ambient noise data in the first storage means is obtained. Thus, pure voice information of the user is obtained, and it is possible to perform voice recognition using this, so that the voice recognition performance can be improved. Furthermore, by applying this voice recognition device, reliable voice dialing can be performed,
This has the effect of enabling security protection.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a voice dial transmission operation of a voice recognition device according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing a voice dial transmission operation of the voice recognition method according to the first embodiment of the present invention.

FIG. 3 is a block diagram showing a voice dial registration operation of the voice recognition method according to the first embodiment of the present invention.

FIG. 4 is a flowchart showing a voice dial registration operation of the voice recognition method according to the first embodiment of the present invention;

FIG. 5 is a configuration diagram of a key telephone system with a voice recognition function according to the first embodiment of the present invention.

FIG. 6 is a configuration diagram of a single telephone with a voice recognition function according to the first embodiment of the present invention.

FIG. 7 is a diagram showing an example of data display on a telephone with a voice recognition function.

FIG. 8 is a block diagram of a conventional voice recognition operation.

FIG. 9 is a flowchart of a conventional voice dialing operation;

FIG. 10 is a flowchart of a conventional voice dial registration operation.

FIG. 11 is a configuration diagram of a conventional key telephone system with a voice recognition function.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Key telephone apparatus 2 Local line interface part 3 Extension interface part 4 System control part 5 Main power supply part 6 Key telephone 7 Child device interface part 8 Communication circuit part 9 Handset 10 Key matrix part 11 LED display part 12 LCD display part 13 Child Machine power supply unit 16 Voice recognition control unit 17 A / D conversion unit 18 Fixed guidance / memory unit 19 Voice telephone directory / memory unit 20 Voice recognition unit 21 Similarity determination unit 22 Standard pattern reproduction unit 23 Control unit 24 Single phone 25 Ambient noise memory Part 26 Body microphone 27 Speaker

Claims (5)

[Claims] 1. A speech recognition for obtaining a difference between an input speech and the ambient noise by storing ambient noise in advance, and searching for and extracting a similar speech from speech information stored in a storage means based on the difference. Method. 2. The speech recognition method according to claim 1, wherein a speech from which ambient noise has been subtracted is stored in advance in the storage means as the speech information. 3. A voice input unit for inputting ambient noise and a user's voice, a first storage unit for storing only the ambient noise, a user's voice input from the voice input unit, and the first voice. Voice extracting means for obtaining a difference from the ambient noise stored in the storing means, a second storing means storing voice information in advance, and a voice extracted by the voice extracting means and the second storing means. A speech recognition device comprising comparison means for comparing and judging stored speech information. 4. The voice extracting means includes third storage means for storing voice input from the voice input means, and
4. A speech recognition apparatus according to claim 3, wherein a difference between the contents of the first and second storage means is obtained. 5. The speech recognition apparatus according to claim 3, wherein a speech obtained from the speech extraction means is used as the speech information of the second storage means.