JP6123121B2 - Voice control system and program - Google Patents

Description

  The present invention relates to a system for controlling the operation of a device by inputting a natural sentence by voice.

  As a conventional example of this type, Japanese Patent Laid-Open No. 2000-56944 is disclosed. According to the invention disclosed in this publication, a natural sentence can be input by voice as described in paragraph 0090. For example, as described in paragraph 0048, “video channel is set to 1”. ”Is entered, as described in paragraph 0051, a command“ turn on power ”and a plurality of commands“ set channel 1 ”can be executed in succession. ing.

  However, natural sentences that can be interpreted by the present invention are limited to requests corresponding to individual buttons of a conventional remote controller, such as “turn on power” and “set channel 1”. For example, “I want to see news” We could not meet the demands of natural sentences.

JP 2000-56944 A

  An object of the present invention is to improve a user's convenience by enabling a speech control system that can execute a plurality of commands by one natural sentence to accept a natural sentence having a wider expression than before.

In order to solve this problem, the present invention is configured as follows.
1. Generates a plurality of continuous control commands for instructing the device to be controlled to perform an action in accordance with the desire according to a request by one natural sentence inputted by voice, and outputs the plurality of continuous control commands to the device to be controlled. In the voice control system,
Conversion auxiliary information for converting a word that is not directly connected to the control command into a parameter for generating the control command, and having a command conversion dictionary that directly connects the word extracted from the natural sentence and the control command. Is stored in the storage means,
A voice control system comprising command conversion means for generating a control command by converting a word that is not directly connected to the control command into a parameter for generating the control command with reference to the conversion auxiliary information.

  According to the present invention, in a voice control system in which a plurality of commands can be executed by one natural sentence, it is possible to accept natural sentences with a wider expression than before, and it is possible to improve user convenience.

The block block diagram which shows 1st Embodiment of this invention. The data structure figure which shows the example of the command conversion dictionary memorize | stored in the memory | storage means of FIG. The block block diagram which shows 2nd Embodiment of this invention.

[First Embodiment]

  The first embodiment of the present invention will be described below. FIG. 1 is a configuration diagram of this embodiment. The voice input means 101 includes a microphone and receives an input of a natural sentence voice. The voice input unit 101 converts the input voice signal into a digital signal and inputs the digital signal to the voice recognition unit 102. The voice recognition unit 102 extracts a plurality of words from the natural sentence voice based on the known voice recognition process, and inputs them to the command conversion unit 103. As a conventional example of extracting a word from a natural sentence by voice recognition processing, for example, there is Japanese Patent No. 3581044. The command conversion means 103 collates a plurality of words extracted by the speech recognition means 102 with a command conversion dictionary, and determines a plurality of commands to be executed corresponding to the natural sentence that is the basis of the words. The conversion dictionary is stored in the command conversion dictionary storage unit 104. The command output means 105 converts the plurality of commands determined to be executed by the command conversion means 103 into signals representing the commands and outputs them to the control target device.

  Further, the command conversion means 103 refers to conversion auxiliary information when determining a command corresponding to a natural sentence. The conversion auxiliary information is auxiliary information for linking a word that cannot be directly linked to the command for the control target device with reference to the command conversion dictionary to the command for the control target device. For example, information of a program guide for deriving a broadcast channel that is currently broadcasting news from the word “news” is included. The conversion auxiliary information is stored in the conversion auxiliary information storage unit 106.

  In addition, the command conversion unit 103 refers to information input from the auxiliary signal input unit 107 when determining a command to be executed corresponding to a natural sentence. The auxiliary signal input unit 107 acquires a signal other than the voice uttered by the user, and includes an appropriate sensor corresponding to the type of the signal. For example, the auxiliary signal input unit 107 includes a camera and photographs the posture and motion of the user's hand. The auxiliary signal input unit 107 converts the captured auxiliary signal into a digital signal and inputs the digital signal to the auxiliary signal recognition unit 108. The auxiliary signal recognition unit 108 recognizes the pattern of information input from the auxiliary signal input unit 107 and inputs the recognized pattern type to the command conversion unit 103. For example, the user's action captured by the camera of the auxiliary signal input unit 107 is recognized as a predetermined gesture, and the recognition result is notified to the command conversion unit 103. For example, Japanese Patent No. 4457983 discloses a conventional example in which a user's motion is captured by a camera, and the motion is recognized and input as a control input.

  A learning instruction unit 109 is connected to the command conversion unit 103. The learning instruction unit 109 includes an operator that can be operated by the user. This operation element is an operation element that is used to control a normal control target device, for example, a button of a remote control for the control target device. When the command conversion unit 103 detects that the operator of the learning instruction unit 109 has been operated after acquiring the words extracted from the natural sentence by the voice recognition unit 102, the command conversion unit 103 and the operated operation The command corresponding to the child is associated and registered in the command conversion dictionary.

  The command output unit 105 is connected to an output destination storage unit 110. The output destination storage unit 110 is set with an identifier of a control target device indicating to which control target device the command should be output. The command output unit 105 converts the command into a signal such as an infrared ray that can be interpreted by the control target device according to the identifier of the control target device set in the output destination storage unit, and outputs the signal. The command output means 105 includes a command signal output device corresponding to the specification of the control target device.

  In FIG. 1, the voice recognition means 102, the command conversion means 103, the command output means 105, and the auxiliary signal recognition means 108 may each be constituted by a dedicated processor, or each means by a general-purpose processor executing a program. You may comprise so that. Moreover, each storage means 104, 106, and 110 should just be provided with the memory area which can memorize | store data, and the kind of element or medium is not limited.

  FIG. 2 is an example of a command conversion dictionary. The command conversion dictionary associates words with commands. In addition, execution priority is associated with the command. In addition, an operator of the corresponding learning instruction unit is associated with the command. A single command can be associated with a plurality of words. For example, a plurality of words such as “ON”, “ON”, and “Turn on” are associated as words corresponding to a power-on command. In addition, the words “TV” and “Watch” are associated with the command “TV power on”, and a single verb selects a specific control target device (TV) from a plurality of control target devices. It is possible to put. For words that are not directly related to the command addressed to the control target device, such as “news”, “music”, and “Music”, the activation of the “channel selection” process performed by the command conversion unit 103 is associated. In addition, a term indicating any one of a plurality of control target devices such as “TV”, “radio”, and “video” is associated with activation of a “device selection” process performed by the command conversion unit 103.

  Priorities are given to each command or process, device selection is No. 1, power-on is No. 2, and channel selection is No. 3. Further, the power button command is associated with the power button operator, and when a power button is operated after a word is input as described above, the word is newly registered in the command conversion dictionary, It is tied to the power-on command. Similarly, a TV power on button is associated with the TV power on command.

  When this system is set to the operating state and the user inputs a natural sentence voice from the voice input unit 101, the voice recognition unit 102 extracts words from the input natural sentence and inputs them to the command conversion unit 103. For example, when “Television is turned on” is pronounced, the words “Television” and “Turn” are recognized and input to the command conversion means 103. The command conversion unit 103 refers to the command conversion dictionary, activates the device selection process with the first priority based on the word “TV”, and sets a specific television as a control target device in the output destination storage unit 110. Next, based on the word “attach”, the command output means 105 is instructed to send a power-on command with the second priority. Based on the setting of the output destination storage unit 110, the command output unit 105 converts the power-on command into a signal that can be interpreted by the television as the control target device, and outputs the signal to the control target device. This turns on the TV.

  For example, when “pronounce news” is pronounced, the words “news” and “see” are recognized and input to the command conversion means 103. The command conversion means 103 refers to the command conversion dictionary and executes a TV power-on command with the second highest priority based on the word “view”. That is, a specific television is set as a control target device in the output destination storage unit 110. Next, the command output means 105 is instructed to send a power-on command. Based on the setting of the output destination storage unit 110, the command output unit 105 converts the power-on command into a signal that can be interpreted by a specific television that is the control target device, and outputs the signal to the control target device. This turns on the TV. Subsequently, the command conversion means 103 executes a “channel selection” process of priority number 3 based on the word “news”.

  That is, the command conversion unit 103 first refers to the conversion auxiliary information and obtains information on the current time stored as the conversion auxiliary information. Next, the TV program table of the day stored as conversion auxiliary information is referred to, and the TV channel on which news is broadcast at the current time is determined. Each program in the program guide is preliminarily embedded with a program type such as news, movie, or soccer. Next, the command conversion means 103 inputs a command for switching the television channel to the channel broadcasting the news to the command output means 105. The command output unit 105 refers to the output destination storage unit 110, converts the command into a signal that can be interpreted by the television as the control target device, and outputs the signal to the control target device. Thereby, the television channel is switched and a news program is displayed.

  The command conversion unit 103 may determine execution of the command based on the input from the auxiliary signal input unit 107. For example, the command is executed only when the user's gesture input from the auxiliary signal input unit 107 matches the predetermined gesture. When there is no predetermined gesture, the control target device is simply picked up the voice of the conversation. It may be determined that there is no intention to control the command, and execution of the command may be stopped. Further, for example, a command to turn on the power may be grasped from voice, and which control target device to turn on may be determined based on a gesture input from the auxiliary signal input unit 107.

  When the control target device is executing a web browser, the command output unit 105 converts the command into a script that can be interpreted by the web browser and outputs the script. In this case, a plurality of commands to be executed continuously are described in one script. When the control target device is on the computer network, the command output means 105 is configured to include a network adapter corresponding to the LAN.

  The voice control system of the present embodiment can be built in a remote controller for a control target device. Further, it can be built in a portable terminal device.

  According to the present embodiment described above, it is possible to perform control such as turning on the television and adjusting the news to the channel being broadcast by the occurrence of one natural sentence by voice. Therefore, in a voice control system in which a plurality of commands can be executed by one natural sentence, it is possible to accept a natural sentence with a wider expression than before, and the convenience of the user can be improved.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. FIG. 3 is a configuration diagram of the present embodiment. In the present embodiment, the voice control system includes the device 100 and the server 200. The basic configuration of the device 100 is almost the same as that of the first embodiment. Unlike the first embodiment, the device 100 includes communication means 111 with the server 200 and the command conversion means 103 transfers the processing to the server when the command conversion means 103 cannot determine a command from words locally. Yes. The operations of the other means are the same as in the first embodiment.

  In this embodiment, the command conversion means 103 of the device 100 requests the server 200 to convert voice to a command when a command corresponding to a word cannot be determined locally. For example, a case where a word not registered in the command conversion dictionary is extracted or a language of another country that is not registered in the command conversion dictionary is extracted. In this case, the command conversion unit 103 acquires from the speech recognition unit 102 an audio file in which the natural sentence audio input to the audio input unit 101 is recorded. Further, the command conversion means 103 acquires from the auxiliary signal recognition means 108 an information file in which information of the auxiliary signal input from the auxiliary signal input means 107 is recorded. Then, the command conversion unit 103 outputs the acquired natural sentence audio file and the auxiliary signal information file to the server 200 via the communication unit 111, and requests the server 200 to convert to a command. The device communication unit 111 and the server communication unit 211 include communication devices compatible with the IP network.

  On the other hand, the server 200 includes the communication unit 211 and the command conversion unit 203. The server 200 also includes command conversion dictionary storage means 204 and conversion auxiliary information storage means 206. The command conversion unit 203 has a function equivalent to that of the command conversion unit 103 of the device 100, and also has a function of the voice recognition unit 102 and a function of the auxiliary signal recognition unit 108. The command conversion dictionary storage unit 204 and the conversion auxiliary information storage unit 206 store a large amount of information that the device 100 does not have in addition to the command conversion dictionary information and conversion auxiliary information that the device 100 has.

  When the device 100 cannot determine the command based on the words and the auxiliary information, the command conversion unit 103 of the device 100 sends the audio file acquired in the device 100 and the information file of the auxiliary information to the server via the communication unit 111. 200. The server 200 receives the audio file and the auxiliary information information file transmitted from the device 100 via the communication unit 211 and accumulates them in the storage unit. The command conversion means 203 of the server 200 refers to the command conversion dictionary and conversion auxiliary information having a larger amount of information than the device 100, and converts the voice and auxiliary information that could not be converted into a command by the device 100 into an appropriate command. To do. Then, the command converted in the server 200 is transmitted via the communication unit 211 and received by the communication unit 111 of the device 100.

  The command conversion unit 103 of the device 100 passes the command received from the server 200 to the command output unit 105. The command output means 105 refers to the output destination set in the local output destination storage means 110, converts the command received from the server 200 into a command signal that can be interpreted by the control target device at the output destination, and Output to the target device. As a result, the operation of the control target device is controlled in accordance with the natural sentence and the auxiliary signal inputted by voice.

  According to the second embodiment, in addition to the effects of the first embodiment, voice and auxiliary information that cannot be converted into a command by the device 100 alone can be converted into a command based on the abundant information amount of the server. In addition, since processing for converting a command into a signal for a control target device is handled locally, private information on what device is specifically controlled can be hidden from the server.

  In the above description, the command converted by the server 200 is configured to be returned to the device 100 that requested the conversion of the command. However, the server 200 addresses the converted command to a device other than the device 100. You may make it transmit. In this case, the IP address of the destination device may be instructed from the device 100 to the server 200.

100 device 101 voice input means 102 voice recognition means 103 command conversion means 104 command conversion dictionary storage means 105 command output means 106 conversion auxiliary information storage means 107 auxiliary signal input means 108 auxiliary signal recognition means 109 learning instruction means 110 output destination storage means 200 Server 203 Command conversion means 204 Command conversion dictionary storage means 206 Conversion auxiliary information storage means 211 Communication means

Claims (7)

In a voice control system for generating a plurality of continuous control commands for instructing a control target device to perform an action in accordance with one natural sentence inputted by voice, and outputting the plurality of continuous control commands to the control target device.
Conversion auxiliary information for converting a word that is not directly connected to the control command into a parameter for generating the control command, and having a command conversion dictionary that directly connects the word extracted from the natural sentence and the control command. Is stored in the storage means,
For words that are not directly linked to the control command, refer to the conversion auxiliary information, convert the control command to a parameter for generating the control command, comprising command conversion means for generating the control command,
The command conversion dictionary registers the plurality of control commands in association with execution priorities,
The voice conversion system, wherein the command conversion means instructs the command output means to output the plurality of consecutive control commands generated based on the one natural sentence according to the priority order. The voice control system according to claim 1.
A voice control system comprising auxiliary signal input means for inputting an auxiliary signal other than voice uttered by a user, wherein the command conversion means generates the control command based on the word and the auxiliary signal. A voice control system according to claim 1 or 2 is provided as a device, and a server that communicates with the device is provided.
The server stores a server-side command conversion dictionary having an information amount exceeding the information amount of the command conversion dictionary stored in the device in the storage unit, and an information amount exceeding the information amount of the conversion auxiliary information stored in the device Server side conversion auxiliary information is stored in the storage means,
When the device cannot generate the control command based on the word by the command conversion dictionary or conversion auxiliary information stored in the device, the device transmits the natural sentence speech information to the server,
The server extracts the word based on the natural sentence audio information received from the device, converts the word into a control command with reference to a server side command conversion dictionary and server side conversion auxiliary information, and the device To
The voice control system, wherein the device outputs a control command received from the server to a control target device. The voice control system according to claim 3.
When the device transmits the audio information to the server, the device also transmits information based on the auxiliary signal input from the auxiliary signal input means to the server,
The voice control system, wherein the server generates the control command based on both the word and information based on the auxiliary signal. The voice control system according to claim 3 or 4,
After receiving a control command from the server, the device converts the control command into a command signal interpretable by the control target device, and outputs the command signal to the control target device. The voice control system according to any one of claims 1 to 5,
When the control target device executes a script, the voice control system describes the plurality of continuous control commands in one script and outputs the script to the control target device. In a voice control system for generating a plurality of continuous control commands for instructing a control target device to perform an action in accordance with one natural sentence inputted by voice, and outputting the plurality of continuous control commands to the control target device.
Conversion auxiliary information for converting a word that is not directly connected to the control command into a parameter for generating the control command, and having a command conversion dictionary that directly connects the word extracted from the natural sentence and the control command. Is stored in the storage means,
The command conversion dictionary registers the plurality of control commands in association with execution priorities,
For a word that is not directly linked to the control command, refer to the conversion auxiliary information, convert the control command to a parameter for generating the control command, and generate the control command;
A process of instructing to output the plurality of consecutive control commands generated based on the one natural sentence according to the priority order;
A voice control program that causes a computer to execute.

Images