JP3116802B2 - Voice recognition method, electronic device, and information storage medium - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a voice recognition method for recognizing input voice, an electronic device, and an information storage medium mounted on the electronic device.

[0002]

2. Description of the Related Art In recent years, various video games for home use have been developed with a variety of exciting games, and some of them are very popular as entertainment. In a video game device, an instruction to switch a game screen or to move a character is given by operating a controller.

On the other hand, as one type of video game device, a video game device for infants mainly for infants has been developed. Some video game devices for infants use picture book type software so that infants can play with interest.When a picture book is designated with a touch pen, the game screen can be switched or displayed on the monitor screen. Various games can be played by designating a specific position on the displayed game screen with a touch pen.

As described above, the conventional video game device is
All instructions to the game were made by the player operating a controller, a touch pen, or the like.

[0005]

The most natural means of communication for humans is voice. For this reason, various attempts have been made to control the machine by human voices since ancient times. Attempts have been made to control vehicles, robots, and the like by human voices, and to control electronic devices by voice instructions.

[0006] In speech recognition technology, adult speech and child speech are completely different, and it is extremely difficult to accurately recognize them using the same speech recognition method. For this reason, a speech recognition mode suitable for each of an adult voice and a child voice is provided. However,
The problem is how to set the optimal speech recognition mode. In the method in which the player selects the voice recognition mode each time, the burden on the player is large, and the recognition may be continued in an incorrect voice recognition mode, or an extra selection operation may detract from the interest of the game. Also, in the case of a video game device for an infant, it is impossible to expect the infant to perform an operation for switching to an optimal voice recognition mode.

An object of the present invention is to provide a voice recognition method and an electronic device that can always perform recognition in an appropriate voice recognition mode. It is another object of the present invention to provide a speech recognition method and an electronic device that can improve the recognition rate of a specific candidate word. It is still another object of the present invention to provide a voice recognition method and an electronic device that can always play a game with an appropriate difficulty level.

Another object of the present invention is to provide an information recording medium suitable for utilizing voice recognition.

[0009]

An object of the present invention is to provide a speech recognition method having a plurality of speech recognition modes. When a predetermined initial condition is detected, the mode automatically returns to the initial speech recognition mode, and a specific operation input is performed. The above is achieved by a voice recognition method characterized in that the mode shifts to a specific voice recognition mode on condition that

The above object is also a speech recognition method having a plurality of speech recognition modes, wherein a state in which a speech recognition result for an input speech continues to be low in one of the plurality of speech recognition modes. Is achieved by a voice recognition method characterized by shifting to another voice recognition mode among the plurality of voice recognition modes. Further, the above object is a speech recognition method for determining a recognition word based on whether or not the similarity of a candidate word to an input speech exceeds a reference value, wherein a plurality of similarities of a specific candidate word to a plurality of input speeches are determined. This is achieved by a speech recognition method, wherein the reference value is determined such that the recognition rate of the specific candidate word becomes a predetermined value based on the degree.

The above object is also a speech recognition method for determining a recognition word based on whether or not the similarity of a candidate word to an input speech exceeds a reference value, wherein the word having a high degree of discrimination from the candidate word is provided. Is added to the candidate word as a dummy candidate word, and when the input voice is recognized as the dummy candidate word, it is determined that the input word cannot be recognized.

In the above-described speech recognition method, a dummy candidate word having a high degree of discrimination with the candidate word has a low probability of being erroneously recognized as the dummy candidate word when the candidate word is uttered, and It is desirable that the word having a low probability of being erroneously recognized as the candidate word when the candidate word is uttered. An object of the present invention is to provide an electronic device that has a plurality of voice recognition modes and recognizes an input voice, a return unit that automatically returns to an initial voice recognition mode when a predetermined initial condition is detected, And a transition unit for transitioning to a specific voice recognition mode.

[0013] In the above electronic device, the predetermined initial condition includes a case where power is input to the electronic device, a case where a program is started, a case where a program is switched, an operation input that deviates from predetermined operation specifications. And / or the case where the similarity of the candidate word to the input speech is low continues.

In the above electronic device, the plurality of voice recognition modes are a child voice recognition mode for a child voice and an adult voice recognition mode for an adult voice. Upon detecting an initial condition, the child recognition mode automatically returns to the child recognition mode,
It is desirable to shift to the adult voice recognition mode on the condition of an operation input combining at least a plurality of operations.

The above object is also achieved by an electronic apparatus having a plurality of voice recognition modes and recognizing an input voice, wherein in one of the plurality of voice recognition modes,
The present invention is achieved by an electronic device including a transition unit that shifts to another voice recognition mode among the plurality of voice recognition modes when the state of a low voice recognition result for the input voice continues.

Further, the above object is achieved by an electronic device for controlling a game by voice input, wherein the difficulty of the game is controlled by changing a recognition rate of the voice input. Further, the object is an information storage medium mounted on the electronic device described above, wherein a sound wave converting means for converting a sound wave into an electric signal, and a processing means for processing the electric signal converted by the sound wave converting means. This is achieved by an information storage medium characterized by having.

In the information storage medium described above, it is preferable that the processing means has a voice recognition means for recognizing a voice from the electric signal converted by the sound wave conversion means.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A video game apparatus according to one embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a perspective view showing the video game device of the present embodiment, and FIG.
FIG. 3 shows a game cartridge of the video game device according to the embodiment, FIG. 3 is a block diagram of the video game device according to the embodiment, and FIG. 4 is an explanatory diagram of a playing method using the video game device according to the embodiment. 5 is a flowchart of basic processing of voice recognition by the video game device of the present embodiment. (Video Game Apparatus Main Body) In the video game apparatus of the present embodiment, a picture book type game cartridge 12 is mounted on a foldable game apparatus main body 10. As shown in FIG. 1, the bottom cover 14 of the game apparatus
6 and open the cartridge slot 18 inside it.
The game cartridge 12 is inserted.

Inside the bottom cover 14 of the game apparatus body 10,
A direction button 20 and an execution button 22 are provided on the left side, a drawing tablet 24 is provided in the center, and a touch pen holder 26 is provided on the right side. The direction button 20 and the execution button 22 are operated by a player. The direction button 20 instructs four directions of up, down, left and right, and the execution button 22 gives an execution instruction. Touch pen holder 2
6 holds the touch pen 28. The touch pen 28
It is used to operate the picture book software by touching the drawing tablet 24. The drawing tablet 24 detects the position where the touch pen 28 is in proximity by the electromagnetic induction method. The position is detected by receiving the electromagnetic signal emitted from the drawing tablet 24 by the touch pen 28.

Inside the upper cover 16 of the game apparatus body 10,
A picture book tablet 32 to which the picture book type game cartridge 12 is mounted is provided at the center. The picture book tablet 32 detects the position where the touch pen 28 is close by the same electromagnetic induction method as the drawing tablet 24. At the lower left part of the picture book tablet 32, a power switch 34 for turning on and off the power is provided. The game cartridge 12 is located on the lower right side of the picture book tablet 32.
Is provided with a cartridge slot 18 into which is inserted. The game cartridge 12 is loaded into the cartridge slot 1 by pressing the eject button 36.
8 can be removed.

On the upper right side of the picture book tablet 32, a page sensor 38 is provided. Game cartridge 1
The page sensor 38 detects which page 42 of the second picture book is open. A cut is formed in the upper right portion of each page 42 of the picture book, and the cut width is increased in the order of pages. By doing so, the number of sensors to be shielded differs depending on the opened page, and the currently opened page can be detected. (Game Cartridge) As shown in FIG. 1, the game cartridge 12 has a basic form like a picture book in which a plurality of pages 42 are bound by a binding ring 40 on the left edge. On each page 42, a picture corresponding to the game program is drawn. As shown in FIG. 1, a microphone 44 for converting sound into an electric signal is attached to a lower left portion of the game cartridge 12.

The details of the game cartridge 12 will be described with reference to FIG. The upper part of the game cartridge 12 is a picture book mounting part 46 for mounting a picture book, and the lower part is a board storage part 50 for storing a circuit board 48 therein. Each page 42 of the picture book is placed on the left edge of the picture book
A binding ring 40 for binding is provided. Six sensor holes 52 corresponding to the positions of the page sensors 38 of the picture book tablet 32 are formed in the upper right portion of the picture book mounting section 46.

The circuit board 48 is housed inside the board housing section 50. On the circuit board 48, in addition to the program ROM 54 storing a game program, the microphone 44 for converting a sound into an electric signal, and a voice recognition LSI 56 for performing voice recognition are mounted. A microphone mounting portion 58 is formed on the surface of the substrate housing portion 50, and the microphone 44 is held by the microphone mounting portion 58.

The microphone 44 is mounted at a position offset from the center of the board storage section 50 of the game cartridge 12. The game cartridge 12 has a standing wave of sound that depends on the shape of the envelope. For this reason, if the microphone 44 is attached to the center of the game cartridge 12, a sound other than the voice will be received in accordance with the antinode of the standing wave. By disposing the microphone 44 at the node of the standing wave at a position shifted from the center of the game cartridge 12, sound other than sound is prevented from entering.

By shifting the position of the game cartridge 12 to the left from the center of the game cartridge 12, when the game cartridge 12 is mounted on the game device body 10, the microphone 44 is positioned at the center of the entire video game device as shown in FIG. Therefore, the player can surely pick up the generated sound. (Configuration of Video Game Apparatus) Next, the configuration of the video game apparatus according to the present embodiment will be described with reference to FIG.

First, the configuration of the game apparatus body 10 will be described. The MPU 60 controls the entire video game device. R
The AM 62 is a work memory of the MPU 60. The video display processor 64 controls a drawing process of a background, a character, and the like. The video information is written to the video RAM 66. The video signal from the video display processor 64 is output to the RGB encoder 68 and output to a television monitor (not shown) via the video output interface 70.

The speech synthesis LSI 72 synthesizes words spoken by a character. Synthesize words according to the game. The mixing circuit 74 mixes the audio signal from the video display processor 64 and the audio signal from the audio synthesizing LSI 72 and outputs the mixed audio signal to a television monitor (not shown) via the audio output interface 76. The button controller 78 monitors the state of the direction button 20, the execution button 22, the touch pen 28, and the page sensor 38, and outputs input information by a player's operation to the MPU 60.

The tablet controller 80 monitors the states of the picture book tablet 32 and the drawing tablet 24 and outputs input information from the operation of the player to the MPU 60. The MPU 60, the RAM 62, the video display processor 64, the voice synthesis LSI 72, the button controller 78, and the tablet controller 80 are connected by an address / data bus 86 and a control bus 88.

Next, the configuration of the game cartridge 12 will be described with reference to FIG. The game cartridge 12 according to the present embodiment has a program RO storing a game program.
Microphone 44 for converting sound into an electric signal together with M54
And a voice recognition LSI 56 (RF5A1
38: manufactured by Ricoh Co., Ltd.). A data register 56a is provided in the voice recognition LSI 56.

Program ROM 54 and LSI for speech recognition
56 is connected by an address / data bus 86 and a control bus 88. The game device body 10 and the game cartridge 12 are connected via connectors 82 and 84, and the address / data bus 86 and the control bus 88 of the game device body 10 and the game cartridge 12 are connected respectively. Note that a power supply line (not shown) is also supplied to the game cartridge 12 via the connectors 82 and 84.

The voice recognition LSI 56 has a volume recognition for identifying the volume of the voice and a word recognition by an unspecified speaker (about 1).
0 words) is possible. Restricting the number of recognized words can improve the recognition rate. When a command is given to the voice recognition LSI 56, the command is written into the data register 56a, and when a recognition result is obtained from the voice recognition LSI 56, the recognition result is read from the data register 56a. (Playing Method of Video Game Apparatus) Next, a playing method of the video game apparatus according to the present embodiment will be described with reference to FIGS.

As shown in FIG. 4, the game device body 10
Then, a television monitor 90 is connected, and the game cartridge 12 provided with the microphone 44 is mounted. The player can select the execution button 22, the direction button 20, the touch pen 28
Instead of, or together with, the user, a voice is emitted toward the microphone 44 to give an instruction to the game.
In the video game device of the present embodiment, volume recognition and voice type recognition are possible. In this specification, volume recognition and voice type recognition are defined as voice recognition together. The volume recognition is a process of recognizing the volume of the voice input to the microphone 44. The type of the voice of an unspecified speaker input to a microphone 44 and the voice type recognition, for example, "Hello", "should", "Tobe", is a process of recognizing such as "Go for it". (Basic processing of speech recognition) The basic processing of speech recognition is shown in the flowchart of FIG. In this specification, "word"
A word does not refer to a word with a strictly grammatical meaning, but one unit for speech recognition, including short ones such as human names and numbers, and long ones such as sentences and sentences. It shall be marked.

First, the recognition command is sent to the data register 56.
Set to a (step S10) Next, the process waits until a recognition result is output from the speech recognition LSI 56 (step S11). Upon receiving the recognition command, the voice recognition LSI 56 stores the recognition result in the data register 56 when the voice having a volume equal to or higher than the predetermined level continues for a certain period of time.
a, and notifies the game machine body 10 that the recognition result has been output.

When the recognition result is output, the first word and the recognition score are read from the data register 56a (step S12). The recognition score indicates the similarity between the input voice and the word of the first recognition by the score. The higher the recognition score is, the more similar the input speech is.
Next, it is determined whether more detailed data is required (step S13), and if not, the process returns to step S10.

If data is required, a next command is set in the data register 56a (step S1).
4). In response to the next command, the voice recognition LSI 56 stores the volume data at the time of occurrence of the recognized word in the data register 56a. The game machine main body 10 reads the volume data from the data register 56a (step S15). Next, it is determined whether more detailed data is necessary (step S1).
6) If no data is needed, return to step S10.

If data is needed, a next command is set in the data register 56a (step S1).
7). In response to the next command, the speech recognition LSI 56 stores the second-ranked word and the recognition score in the data register 56a. The game machine main body 10 reads out the second-ranked word and the recognition score from the data register 56a (step S1).
8).

Next, it is determined whether more detailed data is required (step S19), and if not, the process returns to step S10. If data is needed, a next command is set in the data register 56a (step S
20). LSI 56 for speech recognition by next command
Stores the second-ranked word and the recognition score in the data register 56a. The game machine main body 10 reads the second-ranked word and the recognition score from the data register 56a (step S2).
1).

In this manner, the first word of the recognition and the recognition score can be obtained from the speech recognition LSI as the recognition result, and more detailed data of the recognition result can be obtained by the next command as needed. (Mode Setting in Speech Recognition) Next, a mode setting in speech recognition will be described with reference to FIGS.

For voice recognition technology, adult voices and child voices are completely different, and it is extremely difficult to accurately recognize them using the same voice recognition method. Therefore, an adult voice recognition mode for recognizing adult voices and a child voice recognition mode for recognizing child voices are provided. By recognizing adult voices in the adult voice recognition mode and recognizing child voices in the child voice recognition mode, an improvement in recognition rate can be expected.

However, when the voice recognition mode is incompatible, the recognition rate is significantly reduced, such as the recognition of adult voices in the child voice recognition mode and the recognition of child voices in the adult voice recognition mode. Resulting in. Therefore, it is important to always set the optimal speech recognition mode.
In the present embodiment, since the video game device is for an infant, it is basically conceivable that the infant plays alone.
Since it is conceivable that an adult plays while showing a sample to an infant, it is necessary to switch the voice recognition mode during the play. Also, even if adults can understand that there are two types of speech recognition modes, one for adults and one for children, it is impossible to expect infants to understand the speech recognition mode and to expect the switching operation to the optimal speech recognition mode. is there. Furthermore, it is necessary to prevent the infant from being in the adult speech recognition mode even if it is randomly operated.

In consideration of such points, in this embodiment, the speech recognition mode is set as follows. First, the default value of the voice recognition mode is a child voice recognition mode, and the adult voice recognition mode is a special mode. In the child voice recognition mode, nothing is displayed on the game screen as shown in FIG.
(B) As shown in the upper right part of the game screen, "Adult"
Display.

The switching operation to the adult voice recognition mode is a specific operation input that does not become too complicated for an adult so that even if the infant performs a random operation, it is not accidentally switched. In the present embodiment, as shown in FIG. 7, by pressing the display of “Adult” written on the upper right portion of the picture book page 42 with the touch pen 28 while pressing the execution button 22, the mode is switched to the adult voice recognition mode. To Thus, it is desirable that the specific operation input be an operation input obtained by combining a plurality of operations.

When a specific operation input as shown in FIG. 7 is made, the MPU 60 functioning as a transition means writes a command to switch to the adult speech recognition mode into the data register 56a of the speech recognition LSI 56, and sends the command to the video display processor 64. On the other hand, a command "adult" is displayed in the upper right portion of the game screen shown in FIG. 6B. The return to the child voice recognition mode is automatically switched when a predetermined initial condition described below is detected instead of a special operation input. (1) When the game apparatus body 10 is powered on. (2) When the game cartridge 12 is mounted. (3) When the program is switched by turning over the picture book page 42. (4) When a program using voice recognition is started.

In addition, in the following cases, the mode is automatically switched to the child voice recognition mode as a predetermined initial condition in consideration of the behavior characteristics of the infant. (5) When an operation input outside the predetermined operation specification is continuously performed. The predetermined operation specification is an operation assumed in advance as a normal operation input to the game. For example, it is assumed that the direction button 20 and the execution button 22 are continuously pressed at random, and the drawing tablet 24 and the picture book tablet 32 are continuously and randomly touched by the touch pen 28. (6) A case where the sound volume is higher than a certain level and the recognition score in voice recognition is low. For example, this is the case when a young child keeps saying lousy words in a loud voice. It is assumed that the recognition score is low because of the adult voice recognition mode.

When the above-mentioned predetermined initial condition is detected, the MPU 60 functioning as a return means is turned on by the voice recognition LSI.
The instruction to switch to the child voice recognition mode is written in the data register 56a of 56, and the "adult" display shown in FIG. 6B is erased from the video display processor 64 as shown in FIG. 6A. . Utilizing that the recognition rate is significantly reduced when the speech recognition mode is not suitable, the recognition may always be performed in an appropriate speech recognition mode. In a certain voice recognition mode, when the volume of the voice is higher than a certain level and the recognition score in the voice recognition continues to be low, the MPU 60 functioning as a shift unit.
Command to shift to another voice recognition mode.

For example, in the adult speech recognition mode,
When the sound volume is higher than a certain level and the state where the recognition score in the voice recognition is low continues, the mode is shifted to the child voice recognition mode. Conversely, in the child voice recognition mode, if the volume is at least a certain level and the recognition score in voice recognition is low, the mode is shifted to the adult voice recognition mode. (Improvement of Recognition Rate in Speech Recognition) Next, a method of improving the recognition rate in speech recognition will be described with reference to FIG. 4, FIG. 6, FIG. 8, and FIG.

The LSI 56 for speech recognition of the present embodiment registers 10 candidate words in advance, and
Some of the candidate words are selected. Although it is possible to select all registered candidate words, the recognition rate decreases when the number of selected candidate words increases, so that the recognition rate is usually improved by reducing the number of words to be recognized. I have to.

A specific example of a game using voice recognition is shown in FIG.
This will be described with reference to FIGS. In this game, when the player speaks toward the microphone 44 as shown in FIG. 4, the characters on the screen react variously according to the contents of the player's speech. When the player and say "Hello", the game device is properly recognized as "Hello",
Character on the screen is to say "Hello", as shown in FIG. 8
Bow as shown in (a). Also, when the player utters “packet” (character name) and the game device correctly recognizes “packet”, the character on the screen utters “hi”, as shown in FIG. Jump up. If you can not recognize the voice of the player as "Hello" or "packet", a character on the screen is not a special operation.

[0049] In the case of such a game it is, until now simply, out of the 10 pieces of candidate words registered, voice recognition and selected as a candidate word the two words of "Hello" and "packet" to be recognized I was trying to improve the rate. However, as a result of earnest study by the inventor of the present application, it has been found that erroneous recognition may be increased simply by setting a recognition target as a candidate word.

When the player utters another word, the game device regards any one of the two candidate words as the first place and the other candidate words as the second place. In such a case, since the recognition score is too low even in the first place of recognition, it is determined that recognition is impossible if the recognition score is less than a predetermined threshold. However, depending on the utterance, the recognition score may exceed a predetermined threshold value even for other words, which results in erroneous recognition.

Therefore, in this embodiment, a word having a high degree of discrimination with the recognition target word is added as a dummy candidate word, and if the input speech is recognized as a dummy candidate word, it is determined that recognition is impossible. . With this, when the player utters another word, if the dummy candidate word is ranked first in recognition, even if the recognition score exceeds a predetermined threshold, the word can be rejected, and the false recognition rate is reduced. be able to.

Even if the threshold for the recognition score of the candidate word is lowered to improve the recognition rate, the input speech that is erroneously recognized is recognized as a dummy candidate word.
The recognition rate can be improved without increasing the false recognition rate. It is desirable that the dummy candidate words have a high degree of discrimination from candidate words that are recognition target words. This is because if the player utters another word that is difficult to identify as a candidate word, the dummy candidate word is suppressed and the candidate word is ranked first in the recognition. As a result, erroneous recognition may not be prevented.

A dummy candidate word having a high degree of discrimination from a candidate word has a low probability of being erroneously recognized as a dummy candidate word when the candidate word is uttered, and being erroneously identified as a candidate word when the dummy candidate word is uttered. Words that have a low probability of being recognized. How to select a dummy candidate word will be described using a specific example of FIG. In the present embodiment, "Hello"
"Do your best""Goodbye""Thankyou""Ike"
Ten words, "tore", "tobe", "pico pico", "packet", and "tilo lymph" are registered. All ten words were selected as candidate words, and each of these ten registered words was uttered by an infant about 70 to about 130 times, and the number of candidate words recognized was counted. FIG. 9 shows the result.

[0054] For example, for "Hello" is, 84
For times of voice input, but has been recognized as properly "Hello" 59 times, it is erroneously recognized twice as a "go", is erroneously recognized once as "Tobe", is erroneously recognized 17 times as "Pikopiko"
One misrecognition of "packet" and one misrecognition of "tilo lymph". Other candidate words have not been misrecognized even once. Therefore, the probability of being erroneously recognized word other than the case "Pikopiko" that say "Hello" is seen to be low.

[0055] However, even with the word other than "Pikopiko", undesirable as a dummy candidate words that there is a high probability of erroneously recognized as "Hello" in reverse when you uttered the word. For example, for "Ganbare" includes, but is not also erroneously recognized once as "Hello" when it is uttered "Do your best" and, if you say "Do your best", has also been erroneously as "Hello" recognition 11 times . Therefore, the "Hello", "Do your best" can not be said to be a large degree of identification. On the other hand, for the "good-bye" also not been erroneously recognized once and if you say "hello,""goodbye", also erroneously recognized once and even if you say "goodbye", "Hello" Not. Therefore,
"Hello" and "Goodbye" can be said to be a large degree of identification.

Thus, the recognition results in FIG. 9 show that the candidate words with circles have a high degree of discrimination from each other. When there are a plurality of candidate words as recognition target words, a common word having a high degree of identification may be set as a dummy candidate word. In the present embodiment, "Hello" and because the "packet" is a recognition target word, the common word "goodbye", "should" identify large degree on each
"Tobe" may be selected as a dummy candidate word. (Control of recognition rate in voice recognition) Next, a method of controlling the recognition rate in voice recognition will be described.

FIG. 10 shows a specific example of a game using voice recognition.
This will be described with reference to FIG. In this game, the character is moved to reach the destination according to instructions of four types of voices, "go", "to rare", "tobe", and "do your best". The character is made to walk by an instruction of “go”.
As shown on the left side of FIG. 10A, when the character is stopped, the player utters “Ike”, and when this utterance is correctly recognized, as shown on the right side of FIG. Starts to walk.

The character is stopped by the instruction of "stop". When the character is walking, as shown on the left side of FIG.
When this utterance is correctly recognized, FIG.
As shown on the right side of (b), the character stops. The character jumps according to the instruction "tobe". As shown on the left side of FIG. 11A, when the character walks and reaches the cliff, the player utters “Tobe” with good timing, and when this utterance is correctly recognized, the right side of FIG. As shown in the figure, the character jumps and jumps over the cliff.

A character who is about to fall off a cliff in response to an instruction of “do your best” is saved. As shown on the left side of FIG. 11 (b), when the character is about to fall off the cliff, the player utters “Ganbare”, and when this utterance is correctly recognized, it is shown on the right side of FIG. 11 (b). The character is saved without falling off the cliff. The player uses four types of words, “ike”, “tore”, “tobe”, and “do your best” depending on the scene, and competes to move the character to reach the destination quickly.

An important point in such a game is the recognition rate of the input voice. If the recognition rate is set to be low, the game becomes too difficult, and the player's sense of achievement is not satisfied and is not interesting. Conversely, if the recognition rate is too high, the game becomes too easy and not interesting. Especially in this game,
Since the instruction to the character is given by voice, the recognition or erroneous recognition with a certain probability makes the game interesting. Therefore, the game developer assembles the configuration of the whole game so that it becomes interesting when the input voice is recognized at a certain recognition rate.

However, the recognition rate of the input voice depends on the way of voice and the voice quality of the player. In addition, the recognition rate greatly differs even for the same player due to ambient noise. For this reason, it is difficult to realize the fun of the game intended by the developer. Therefore, in this embodiment, the reference value for speech recognition is adaptively changed based on the input speech so that the recognition rate of the input speech becomes a predetermined value. If the speech recognition method determines the recognition word based on whether or not the similarity of the candidate word to the input voice exceeds a reference value, based on the recognition score of a specific candidate word for a predetermined number of input voices so far, The reference value is determined so that the recognition rate of a specific candidate word becomes a predetermined value. For example, if you want to increase the recognition rate to 80%,
The reference value is changed so that eight recognition scores out of zero recognition scores are exceeded.

The lower the recognition score is, the lower the reference value is. The higher the recognition score is, the higher the reference value is. However, from the viewpoint of the player, the recognition rate is stable, so that even if the player changes or the surrounding environment changes, You can always enjoy the game with appropriate difficulty. Further, the difficulty of the game can be changed by changing the recognition rate of voice recognition.

When the recognition rate of the voice recognition is low, the player must be careful to pronounce the sound, and the probability that the player will not be correctly recognized increases, thereby increasing the difficulty of the game. Using this, the difficulty level of the game can be controlled intentionally. For example, when the character reaches the destination in the games shown in FIG. 10 and FIG. 11, when trying again, the reference value for voice recognition is increased to increase the difficulty of the game. Conventionally, in order to increase the difficulty of the game, the speed is increased, the number of enemies is increased, or the like.However, according to the present embodiment, the reference value for determining the recognition rate of voice recognition is changed. By doing so, the difficulty of the game can be easily controlled. (Modified Embodiment) The present invention is not limited to the above embodiment, and various modifications are possible.

For example, in the above embodiment, words such as "ike" and "tobe" are targeted for recognition. However, the recognition target is not limited to words, but may be voice such as a person's name or a sentence. In the above-described embodiment, the present invention is applied to the game cartridge mounted on the video game apparatus. However, any information storage medium mounted on the electronic apparatus may be used. For example, an information storage medium such as a memory card may be used.

In the above-described embodiment, the present invention is applied to the video game device. However, any electronic device may be used as long as the information storage medium is mounted. For example, electronic devices such as an electronic organizer, a personal computer, a calculator, a facsimile, and a printer may be used.

[0066]

As described above, according to the present invention, when a predetermined initial condition is detected, the mode automatically returns to the initial voice recognition mode, and shifts to the specific voice recognition mode on the condition of a specific operation input. , The recognition can always be performed in an appropriate voice recognition mode. Further, according to the present invention, in a voice recognition mode among a plurality of voice recognition modes,
If the speech recognition result for the input speech continues to be low, the mode shifts to another speech recognition mode, so that the speech can always be recognized in an appropriate speech recognition mode.

Further, according to the present invention, the reference value is determined such that the recognition rate of the specific candidate word becomes a predetermined value based on the plurality of similarities of the specific candidate word with respect to the plurality of input voices. Therefore, the player can always enjoy the game with an appropriate difficulty level. According to the present invention, a word having a high degree of discrimination from a candidate word is added to the candidate word as a dummy candidate word, and when the input voice is recognized as a dummy candidate word,
Since it is determined that recognition is not possible, the erroneous recognition rate of the recognition target word can be reduced.

Further, according to the present invention, the difficulty of the game is controlled by changing the recognition rate of the voice input, so that the difficulty of the game can be easily controlled.

[Brief description of the drawings]

FIG. 1 is a perspective view showing a video game device according to an embodiment of the present invention.

FIG. 2 is a view showing a game cartridge of the video game device according to one embodiment of the present invention.

FIG. 3 is a block diagram of a video game device according to an embodiment of the present invention.

FIG. 4 is an explanatory diagram of a playing method of the video game device according to one embodiment of the present invention.

FIG. 5 is a flowchart showing basic processing of voice recognition in the video game device according to one embodiment of the present invention.

FIG. 6 is a diagram showing a game screen of a specific example of a game according to an embodiment of the present invention.

FIG. 7 is a diagram showing an operation of switching to an adult voice recognition mode in one embodiment of the present invention.

FIG. 8 is a diagram showing a game screen of a specific example of a game according to an embodiment of the present invention.

FIG. 9 is a diagram showing a recognition result of voice recognition according to an embodiment of the present invention.

FIG. 10 is a diagram showing a game screen of another specific example of the game according to the embodiment of the present invention.

FIG. 11 is a diagram showing a game screen of another specific example of the game according to the embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Game apparatus main body 12 ... Game cartridge 14 ... Bottom cover 16 ... Top cover 18 ... Cartridge slot 20 ... Direction button 22 ... Execute button 24 ... Drawing tablet 26 ... Touch pen holder 28 ... Touch pen 32 ... Picture book tablet 36 ... Eject button 38 ... Page sensor 40 Ring 42 Page 44 Microphone 44b Rubber holder 46 Picture book mounting section 48 Circuit board 50 Board storage section 52 Sensor hole 54 Program ROM 56 Voice recognition LSI 58 Microphone mounting section 60 ... MPU 62 RAM 64 Video display processor 66 Video RAM 68 RGB encoder 70 Video output interface 72 Audio synthesis LSI 76 Audio output interface 74 Mixing circuit 78 Button controller 80 ... Tablet controller 86 ... Address / data bus 88 ... Control bus 90 ... TV monitor

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-57-160487 (JP, A) JP-A-7-104786 (JP, A) JP-A-59-228880 (JP, A) JP-A-59-228880 48799 (JP, A) JP-A-60-237532 (JP, A) JP-A-62-39899 (JP, A) JP-A-58-23097 (JP, A) JP-A-59-2359 (JP, A) JP-A-57-202597 (JP, A) JP 4-45839 (JP, B2) JP-B 61-22320 (JP, B2) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 15/00-17/00

Claims (6)

(57) [Claims] An electronic device that has a child voice recognition mode for a child voice and an adult voice recognition mode for an adult voice and recognizes an input voice. The electronic device further comprises a child voice recognition mode transition unit that automatically shifts to the child voice recognition mode when detecting the child voice recognition mode. 2. The electronic device according to claim 1, further comprising: an adult voice recognition mode shift unit that automatically shifts to the adult voice recognition mode when a specific operation input is detected in the child voice recognition mode. An electronic device, comprising: 3. The electronic device according to claim 2, wherein the specific operation input is an operation input obtained by combining a plurality of operations. 4. The electronic device according to claim 1, wherein the predetermined initial condition includes: when power is input to the electronic device; when a program is started; An electronic device, wherein the switching is performed, the operation input that deviates from a predetermined operation specification is continuously performed, and / or the case where the similarity of the candidate word to the input voice is low continues. 5. The electronic device according to claim 1, further comprising a display for displaying a current voice recognition mode. 6. An electronic device for controlling a game by voice input, wherein the difficulty of the game is controlled by changing a recognition rate of the voice input.