JPH0225896A - voice recognition device - Google Patents

Abstract

PURPOSE:To cause plural persons to use one device by providing a display part and displaying the kind of standard patterns used for a recognition or the kind of storing parts to store it. CONSTITUTION:A dictionary for an unspecific speaker is registered to a register 5 beforehand, and a dictionary for a specific speaker to be freely registered by a user is in a register 6. The internal part of the register 6 is divided into several parts, several users use them separately, and it is sufficient for the user to designate the part of the dictionary of himself at the time of using. In order to register it, a switch 4 is made to fall at a B side, and it can be used for the unspecific speaker and for the specific speaker when a switch 7 is made to fall at a C side and a D side next, respectively. For example, when the switch 7 falls at the C side, the fact that the dictionary for the living unspecific speaker is used is displayed on a display part 8. Thus, one device can be used by plural persons.

Description

[Detailed description of the invention] Watched for 11 minutes The present invention relates to a speech recognition device.

Research on voice recognition is currently active, but there are two types of speech recognition: the speaker-specific method, in which the user registers his or her own voice, and the unspecified-speaker method, which can be used without any preparation. There is,
The former has a higher recognition rate than the latter. On the other hand, instead of using these as separate machines, it is conceivable to have both functions in one and use them properly as needed.

In this case, commonly used words are recognized using a speaker-independent method,
If this is not the case, it would be convenient to recognize it using the speaker-specific method. Alternatively, you can leave the command to specify a specific specification unspecified, for example, if you say your name,
It is convenient to use it in such a way that the person's speech dictionary is searched from the disk and loaded. In such cases, one device is used by several people, and if one person gets up while using it, the voice dictionary of the previous person will be loaded when the next person uses it. It has the disadvantage that it is not recognized.

-15 The present invention was made in view of the above-mentioned circumstances,
In particular, when a single recognition device is used by multiple people, the purpose of this invention is to provide a device that can be used by any person at any time.

I'm in love with you.

In order to achieve the above object, the present invention provides a means for converting audio into an electrical signal, a feature converter for extracting a feature from this signal, a storage for storing the feature, and a different controller. 2, a comparison unit for comparing feature amounts, and a display unit for displaying information related to recognition; the type of standard pattern used for recognition or the type thereof is stored This device is characterized in that the type of storage section is displayed on the display section. The following will explain based on five embodiments of the present invention.

FIG. 1 is a block diagram for explaining one embodiment of the present invention. In the figure, 1 is a microphone, 2 is a microphone amplifier, 3 is a group of band pass filters, 4 is a switch, 5 and 6 are registers, and 7 is a block diagram for explaining an embodiment of the present invention. 8 is a switch, 8 is a display section, 9 is a comparison section, and 10 is a result output section. Microphone 1 is used as an acousto-electrical transducer, and microphone amplifier 2 is used as an amplifier to amplify the electric signal from the microphone. It is input to band pass filter group 3 and frequency analyzed. At this time, the feature quantity is the power spectrum, but it is not limited to this, and the LPG
However, the number of zero crossings is also fine. Further, a dictionary for unspecified speakers is registered in advance in the register 5, and a dictionary for specific speakers, which can be freely registered by the user, is stored in the register 6. The register 6 is divided into several sections, which are used by several users, and all they have to do is specify their own section of the dictionary when using it. To register this, turn the switch 4 to the B side. When the switch 7 is turned to the C side, it can be used for unspecified speakers, and when it is turned to the 5D side, it can be used for a specific speaker. When the switch 7 is turned to the C side, it is displayed on the cathode ray tube of the display section 8 that the dictionary for unspecified speakers is currently being used. The display method may be one that indicates that the dictionary currently in use is for general use, as shown in Figure 2, or one that indicates that the dictionary for specific speakers is divided into several sections. When using the A part of the dictionary, the letter "A" may be reversed as shown in FIG. 3. In any case, the dictionary currently being used should be clearly indicated. In addition, when in use, the switch 4 is turned to the A side, and the comparison section 9 compares the input voice pattern with the pattern in the register to find the degree of similarity, and recognizes it based on the magnitude of the similarity. It is not limited. A method in which the specific speaker method and the non-specific speaker method can be distinguished only by the structure of the dictionary pattern (Mathematical Science J No. 287 February 1987 issue P, P, 6
3-69. Fuzzy speech recognition) etc. are desirable.

Further, as shown in FIG. 4, it is desirable to provide a switch 11 so that when the switch 11 is pressed, the switch 7 falls to the C side, that is, to the side of the dictionary for unspecified speakers. Insert a command (for example, your name) to specify your own dictionary (for specific speakers) in the dictionary for non-specific speakers, and when you say your name to the device, it will automatically display your specific dictionary. A dictionary for the speaker is loaded. In this state, if you use it as a device for a specific speaker and leave your seat while it is still in use, the next user will be able to see at a glance whose dictionary is currently loaded, and if it is not his/her dictionary, press switch 11. You will be able to use it immediately.

As is clear from the above description, the present invention allows one device to be used by multiple people and eliminates the hassle of having to check whose dictionary is loaded. To be released.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram for explaining an embodiment of the present invention; FIGS. 2 and 3 are diagrams showing display examples of the display unit;
The figure is a configuration diagram for explaining another embodiment of the present invention. 1...Microphone, 2...Mic amplifier, 3.
... Bandpass filter group, 4... Switch, 5, 6
...Register, 7...Switch, 8...Display section,
9... Comparison section, 10... Result output section, 11... Switch. Ward chart chart chart chart chart

Claims (1)

[Claims] 1. A means for converting audio into an electrical signal, a feature converter that extracts a feature from this signal, a storage that stores the feature, a second storage that is different from this, and a comparison of the feature. In a speech recognition device that has a comparison section that performs the following operations, and a display section that displays information related to recognition, the type of standard pattern used for recognition or the type of storage section that stores it is displayed on the display section. A speech recognition device characterized by: