JP2002333898A - Sound-recognizing system for electronic pet - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform efficient sound recognition without misrecognitions especially by comparing the difference in the recognition rates of respective candidates. SOLUTION: In the sound recognizing system for electronic pet, using an electronic pet such as human faced fish 'Seaman', in terminal equipment for storing the program of the electronic pet, in the case of conversation between the 'Seaman' and an operator, the voice of the answer of the operator is recognized and a plurality of candidates having different recognition rates are presented. When there is a difference of recognition rates more than a prescribed value between first and second candidates, the first candidate is defined as being recognized result. When there is no difference of recognition rates more than the prescribed value between the first and second candidates, confirmation processing is applied on the first candidate, the misrecognition is eliminated, and efficient sound recognition is enabled.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an electronic pet voice recognition system and method.

[0002]

2. Description of the Related Art With the recent advance in electronics technology, there is an electronic pet program that accumulates information regarding the personality of an operator as knowledge and realizes a response suitable for the operator based on the accumulated various data. For example, there is a system that talks with an electronic pet swimming in a water tank and communicates with an operator.

[0003]

In such a communication system with an electronic pet, communication proceeds by conversation between the electronic pet and a user. At this time, voice recognition and voice synthesis of the electronic pet and the user are performed using a voice recognition engine and a voice synthesis engine.

However, in the conventional speech recognition, it cannot be said that reliable recognition is always performed. For example, when recognizing the answer to the question of the mermaid "Seaman", the recognition result with the highest recognition rate was used as the answer. For this reason, even when the recognition rate itself is not so high, it is determined that the answer is given, and there is a risk of erroneous recognition.

[0005] Some systems always include a process for confirming the recognition result to the user. However, this is a very inefficient speech recognition method. The present invention has been made in view of the above problems, and has as its object to perform more accurate voice recognition by comparing differences in recognition rates.

[0006]

According to the first aspect of the present invention, there is provided an electronic apparatus comprising: a plurality of terminal devices; and a server computer connected via a network and connected to enable data communication. A voice recognition system for a pet, comprising: voice recognition means for outputting a plurality of candidates having different recognition rates for an electronic pet question; and a recognition rate between a first candidate and a second candidate obtained by the voice recognition. Comparing means for comparing the difference between the first candidate and the speech recognition result when the comparison result by the comparing means exceeds a predetermined value. If not, it can be achieved by providing a voice recognition system for an electronic pet having a confirmation means for performing a confirmation process on the first candidate.

With this configuration, when there is a large difference between the first candidate and the second candidate in the recognition rate, the first candidate is determined to be a correct answer and speech recognition is performed, and efficient speech recognition processing is performed. Is what you do.

According to a second aspect of the present invention, in the invention according to the first aspect, when a positive answer is received in the confirmation processing for the first candidate, the first candidate is recognized as the result of the speech recognition. It is a configuration to do.

With this configuration, the first candidate can be determined as a result of speech recognition, and erroneous recognition can be prevented. According to a third aspect of the present invention, in the invention according to the first aspect, in the confirmation processing for the first candidate, if a negative response is received, the confirmation processing for the second candidate is further performed.

[0010] With this configuration, it is possible to efficiently perform the speech recognition processing for the second candidate and the subsequent candidates without erroneous recognition. According to a fourth aspect of the present invention, in the configuration of the third aspect, when a positive response is received in the confirmation processing for the second candidate, the second candidate is recognized as the result of the speech recognition. is there.

With this configuration, the second
Speech recognition processing can be performed efficiently for candidates and below without erroneous recognition. According to a fifth aspect of the present invention, in the third aspect, when a negative response is received in the confirmation processing for the second candidate, the confirmation processing for the third candidate is performed.

According to a sixth aspect of the present invention, in the first to fifth aspects, the recognition rate of the first candidate is a certain value or more. With such a configuration, for example, when the first candidate has a low recognition rate and a difference between the second candidate and the second candidate is equal to or more than a predetermined value, the first candidate having the low recognition rate Is prevented from being determined.

[0013] The above object is attained according to a seventh aspect of the present invention.
A plurality of terminal devices, a voice recognition method of an electronic pet including a server computer connected via a network network and communicably connected, the processing for voice recognition of the answer to the question of the electronic pet, When a plurality of recognition candidates are obtained by the voice recognition, a comparison process of comparing the difference in the recognition rate between the first candidate and the second candidate, and when the comparison result by the process exceeds a predetermined value, A voice recognition method for an electronic pet, comprising: a certification process for recognizing a first candidate as a result of the voice recognition; and a confirmation process for performing a confirmation process on the first candidate when a comparison result by the process does not exceed a predetermined value. Can be achieved by providing

The present invention is an invention of a method. With such a configuration, when there is a large difference between the first candidate and the second candidate in the recognition rate, the first candidate is determined to be a correct answer, Can recognize voice.

According to an eighth aspect of the present invention, in the above-mentioned seventh aspect, the recognition rate of the first candidate is equal to or more than a predetermined value. According to the ninth aspect of the present invention, there is provided a voice recognition program for an electronic pet, comprising: a plurality of terminal devices; and a server computer connected via a network and connected to enable data communication. A function of recognizing the answer to the question of the electronic pet by voice, and a comparing function of comparing a difference in recognition rate between the first candidate and the second candidate when a plurality of recognition candidates are obtained by the voice recognition; A certifying function for certifying the first candidate as a result of the speech recognition when the comparison result by the comparison function exceeds a predetermined value; and a certifying function when the comparison result by the function does not exceed a predetermined value.
This can be achieved by providing a voice recognition program for an electronic pet having a confirmation function of performing a confirmation process for a candidate.

The present invention is an invention of a program. With this configuration, when there is a large difference in recognition rate between the first candidate and the second candidate, the first candidate is determined to be a correct answer, Can recognize voice.

According to a tenth aspect, in the invention of the ninth aspect, the recognition rate of the first candidate is equal to or more than a predetermined value.

[0018]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a system configuration diagram of a voice recognition system using an electronic pet according to the present invention. As shown in FIG. 1, the system of the present embodiment includes a plurality of personal computers 1a,
And 1b, a computer game machine 6, and a mobile phone 7
And a server computer 3 owned by a service provider that provides a service for providing a mermaid "Seaman" as an electronic pet.

The personal computers 1a, 1b,
The game machine 6 is connected to the Internet 21 via each of the communication devices 2a, 2b, and 5, and the mobile phone 7 is connected to the Internet 21 via a not-shown relay station or central control station.

Although personal computers connected to the Internet 21 are indicated by 1a and 1b, many computers are connected to the Internet 21 in addition to the above two computers. The personal computer 1a uses, for example, Windows as the OS,
The personal computer 1b uses, for example, a Macintosh as the OS.

First, the configuration of the personal computer 1 used by the receiver and the sender in this embodiment is shown in FIG. In the description of the configuration of the personal computer 1, the personal computer 1a will be representatively described.

As shown in FIG. 2, a data bus 10 for transmitting and receiving data is provided inside the personal computer 1a.
The data bus 10 is connected to a central processing unit (hereinafter, referred to as a CPU) 11, a RAM 12, a real-time clock (RTC) 17, and the like.

The CPU 11 outputs a question to the user and accepts an answer, which will be described later, as words of an electronic pet displayed on the display device by speech synthesis, and recognizes and pre-registers the answer of the user by voice recognition. A response receiving process for selecting and registering a corresponding item from a plurality of answer candidates set as above and storing the selected item as personality information of the user is performed. During this process,
The RAM 12 is used as a work area.

A real-time clock (RTC) 1
Reference numeral 7 outputs, for example, current time information used for date and time information of data update, and calendar information such as an arbitrary day of the week. The input device 16 is a keyboard, a mouse, and the like.
Notify.

The display device 14 is a display such as a CRT or an LCD, and displays a Seaman described later. The communication interface 13 is connected to the communication device 2 (2
a) and transmits / receives data to / from the server computer 3 via the communication device 2 (2a) and the Internet 21.

The audio input / output device 18 is connected to a speaker 19 and a microphone 20 as external devices, and A / D / D
/ A converter. The voice input / output device 18 converts (D / A converts) voice data generated by a voice synthesis program to be described later into analog voice, outputs the voice data to the speaker 19, and converts the voice input from the microphone 20 into a digital voice. The data is converted (A / D converted) and output.

The storage device 15 is composed of a magnetic disk or a magneto-optical disk. Data and programs are written and read under the control of the CPU 11. The storage device 15 stores an electronic pet program in which an image and an operation process of a mermaid fish "Seaman", which is an electronic pet displayed on the display device 14, and a voice of the mermaid fish "Seaman" are stored. A speech synthesis program for outputting speech based on text data, a speech recognition program for converting input speech of a sender or a recipient into text data, and the like are stored.

In the storage device 15, a group of words used and understood by the mermaid "Seaman" in the application is registered as a "dictionary". FIG. 3 is a diagram showing the structure of this dictionary, which is registered for each category.

For example, in item number 1, the question "You are a man" of the mermaid "Seaman" and the answers a and b to the question are registered. In item number 2, a question "Your age" of the mermaid "Seaman" and answers a, b,... To the question are registered. Hereinafter, as shown in the figure, the question of the mermaid "Seaman" and its answer are registered as a multi-word dictionary. And
At the time of a recognition process described later, a comparison process is performed for each word for the user's answer, and a recognition rate is output for each word.

The processing operation of the above-structured speech recognition system will be described below. FIG. 4 is a flowchart illustrating the processing operation of this example. First, a question is asked from the mermaid "Seaman" (step (hereinafter referred to as ST) 1
Is YES). This question is made in the order of the data stored in the above-mentioned database, and this question flows from the speaker 19a of the personal computer 1a (ST2). For example, if the question asked by the mermaid "Seaman" is the item number 2 shown in FIG. This question flows from the speaker 19a, and the user understands the question.

Next, the user answers the above-mentioned question and answers the microphone 20a (YES in ST3). CP
U11 performs voice recognition from the above answer and selects a plurality of candidates (ST4). For example, the user's answer is “27 years old”
, The candidate shown in FIG. 5 is selected. That is,
The answer in this case is answer a corresponding to item number 2 above,
are selected from b, c,..., and are output in descending order of recognition rate based on a speech comparison for each word, for example, a comparison of integration values. For example, as shown in FIG. 5, "Juneha" is output as a first candidate, "Junea" as a second candidate, "Junea" as a third candidate, and so on.

At the same time, the recognition rate of each candidate is also displayed. For example, the first candidate is 60% and the second candidate is 55%.
%, And the third candidate is 45%. Next, the CPU 11
Determines whether the difference between the recognition rates of the first candidate and the second candidate is n points or less (ST5). The n points are set in advance, and are values that do not cause erroneous recognition, for example, through experience.
Here, for example, if the n point is set to “20”, the result is YES in the example of FIG.

As described above, if the difference between the recognition rates of the first candidate and the second candidate is n points or less (YES in ST5), the first candidate is presented to the user (ST6). That is, in this case, since the recognition result is not reliable, the user is urged to confirm. For example, in the above example, the user is prompted to confirm that he is now 17 years old (see FIG. 6). On the other hand, the aforementioned judgment (ST
In 5), if the difference between the first candidate and the second candidate is n points or more (NO in ST5), the first candidate is determined, the answer is obtained, and the above-described confirmation processing is not performed. The face fish “Seaman” says “17 years old” from the speaker 19a (see FIG. 7).

Next, the first candidate confirmation processing (ST6)
As a result of performing the notification from the speaker 19a and the display shown in FIG. 6, if there is a positive answer from the user, also in this case, the first candidate is determined as the answer (ST8 is YE).
S, ST9). For example, when the user returns an answer such as “Yes”, “Yes”, “Yes”, etc., toward the microphone 20a, the first candidate is determined as the answer.

On the other hand, if the user makes a negative answer such as "No" or "No", the mermaid "Seaman" determines whether there is a next candidate, and if there is a second candidate, The candidates are presented (YES in ST10, ST11). For example,
In the case of the example shown in FIG. 5, the second candidate “Ninare” is presented. In other words, the mermaid "Seaman" asks a question, "Well, I said 27 years old."

Thereafter, the same processing is performed, and if there is a positive answer to the presentation of the second and subsequent candidates, the answer is regarded as answer acquisition. For example, if the user has a positive answer to the second candidate presentation at the age of 27, the user is known to be 27 years old, and a positive answer is to the third candidate presentation at the age of 57. If so, it can be understood that the age of the user is 57 years old.

By repeating the above processing, the mermaid "Seaman" can know the correct age of the user. If all candidates are denied by the user,
Return the conversation to the first question.

By performing the above control, the recognition rate of speech recognition is output in response to the question of the mermaid "Seaman", and the speech recognition process is performed based on the recognition rate, and the user can be stress-free. A dialogue is realized.

Although the question about the mermaid "Seaman" was a question about the age for the user, accurate information from the user can be obtained by the same processing for a question other than the age.

When the recognition rate of the first candidate is equal to or less than a predetermined value, even if there is a difference in the recognition rate between the first candidate and the second candidate equal to or more than a predetermined value, the recognition result is not obtained as a speech recognition result. In such a case, a procedure such as performing a confirmation process for both candidates is performed to provide a system that does not cause erroneous recognition.

In the above processing, the terminal device used is the personal computer 1a, but may be another personal computer 1b or the game machine 6b.
Or the mobile phone 7.

The personal computer or the like may not be connected to the server, and may have a configuration in which the corresponding function is independently held by a circuit or a program.

[0043]

As described above, according to the electronic pet voice recognition system of the present invention, erroneous recognition is eliminated and efficient voice recognition is enabled.

Further, when the recognition rate of the first candidate is equal to or less than a certain value, a more accurate voice recognition system can be realized by performing processing such as confirmation processing for both candidates.

[Brief description of the drawings]

FIG. 1 is a system diagram of an electronic pet voice recognition system of the present invention.

FIG. 2 is a block diagram showing a computer owned by a user used in the embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of a database registered in a storage device.

FIG. 4 is a flowchart illustrating a processing operation in the present example.

FIG. 5 is a diagram showing recognition rates of a first candidate, a second candidate, and a third candidate.

FIG. 6 is a diagram showing a sounding state of a mermaid “Seaman”.

FIG. 7 is a diagram showing a sounding state of a mermaid “Seaman”;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Personal computer (user) 2 Communication device 3 Server computer 4 Communication device 5 Communication device 6 Computer game machine 7 Cellular phone 8 Television 9 Microphone 10 Data bus 11 Central payment processing device (CPU) 12 RAM 13 Communication interface 14 Display device 15 Storage device 19 Speaker 20 Microphone 21 Internet network

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G10L 3/00 571U 561E F-term (Reference) 2C001 AA00 AA11 BA00 BA06 BA07 CA00 CA07 CB01 CB04 CB08 CC01 CC08 5D015 KK02 LL04 LL05

Claims (10)

[Claims] 1. An electronic pet voice recognition system comprising a plurality of terminal devices and a server computer connected via a network and connected to enable data communication, wherein the voice recognition system recognizes a question from the electronic pet. Speech recognition means for outputting a plurality of candidates having different rates, comparison means for comparing the difference in recognition rate between a first candidate and a second candidate obtained by the speech recognition, and a comparison result by the comparison means being a predetermined value When the value exceeds the value, the certifying means for certifying the first candidate as the result of the speech recognition, and when the comparison result by the comparing means does not exceed a predetermined value,
And a confirmation means for performing a confirmation process on the first candidate. 2. The electronic pet according to claim 1, wherein in the confirmation processing for the first candidate, when a positive response is received, the second candidate is recognized as a result of the voice recognition. Voice recognition system. 3. The voice recognition system for an electronic pet according to claim 1, wherein in the confirmation processing for the first candidate, if a negative response is received, the confirmation processing for the second candidate is further performed. . 4. The electronic pet according to claim 3, wherein in the confirmation processing for the second candidate, if a positive response is received, the second candidate is recognized as the result of the voice recognition. Voice recognition system. 5. The voice recognition system for an electronic pet according to claim 3, wherein a confirmation process is performed on the third candidate when a negative response is received in the confirmation process on the second candidate. 6. The voice recognition system for an electronic pet according to claim 1, wherein the recognition rate of the first candidate is equal to or higher than a predetermined value. 7. A method for recognizing a voice of an electronic pet, comprising a plurality of terminal devices and a server computer connected via a network and connected to enable data communication, wherein a response to an electronic pet question is voiced. Recognition processing; and when a plurality of recognition candidates are obtained by the voice recognition, a comparison processing of comparing a difference in recognition rate between the first candidate and the second candidate; When exceeding, the first
A certification process for recognizing the candidate as a result of the speech recognition; and a confirmation process for performing a confirmation process on the first candidate when a comparison result by the process does not exceed a predetermined value. Pet voice recognition method. 8. The method according to claim 7, wherein the recognition rate of the first candidate is equal to or higher than a predetermined value. 9. A voice recognition program for an electronic pet, comprising: a plurality of terminal devices; and a server computer connected via a network and connected to enable data communication, wherein a voice response to a question of the electronic pet is provided. A recognition function; a plurality of recognition candidates obtained by the voice recognition; a comparison function of comparing a difference in recognition rate between the first candidate and the second candidate; and a comparison result obtained by the comparison function being a predetermined value. A certification function for certifying the first candidate as a result of the speech recognition when the number exceeds the threshold, and a confirmation function for performing a confirmation process on the first candidate when a comparison result by the function does not exceed a predetermined value. A voice recognition program for an electronic pet, comprising: 10. The computer-readable storage medium according to claim 9, wherein a recognition rate of the first candidate is equal to or greater than a predetermined value.