JP2017191390A - Communication system, communication log collection system, server, and communication method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase the quality of communications between a user and a robot.SOLUTION: The present invention relates to a communication system having a robot on a user side and a server. The robot includes: a microphone for collecting sounds of a user; sending/receiving means for sending a sound signal of a sound of a user collected by the microphone to the server via a network and receiving a sound signal of the user sent via the network; sound editing means for editing the sound signal of the user sent via the network as a speech of the robot and generating an edited sound signal; and at least one speaker for outputting the edited sound signal. The server has connection management means for managing the sending and receiving of signals of the robot.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a communication system, a conversation log collection system, a server, and a communication method.

  In recent years, a communication system that establishes a dialogue between a person and a robot (for example, a humanoid robot) has been proposed.

  One of them is called a task-oriented type, which is an interactive system that allows a robot to perform a specific task. For example, in response to an utterance (command) of a person who is a user such as “tell me today's weather”, the robot conveys today's weather forecast by voice. A set of these commands and answers is uniquely registered in advance in the dictionary.

  The other is called a chat type, which is a system that allows a user to enjoy a conversation with a robot rather than letting a task unique to the robot (Non-Patent Document 1). This applies a chatbot (artificial incompetence) dialogue system. This chatbot dialogue system is a system for carrying out daily conversation with a user, and is roughly classified into a dictionary type (scenario type), a log type, a Markov sentence generation type (text generation type), and the like. The basic point is that a predetermined dialogue pattern is stored in a database, a suitable response content is searched according to the input content at the time of dialogue, and it is output from the system side. For example, when the user inputs “What do you like?” To the interactive system through a keyboard, microphone, etc., the system side will respond most closely to a word string such as “What is it? Search for. A large number of input examples and corresponding response sentences are stored in the database in advance. The dialogue system takes out a response sentence selected according to the search result and outputs it to the user via a speaker or a monitor. By devising a method for storing response contents in the database, it is possible to insert part of the user's input into the response sentence.

[Valerie] Valerie Web Site: http://www.roboceptionist.com/

  However, at present, none of the methods such as dictionary type (scenario type), log type, Markov sentence generation type (text generation type) are perfect, and there are many cases where a conversation between a human and a robot is not established. .

  On the other hand, the user of the robot desires communication completed to some extent with the robot.

  Therefore, the present invention has been invented in view of the above problems, and its purpose is a communication system, a conversation log collection system, a server, and a communication method that can improve the quality of conversation performed between a user and a robot. Is to provide.

  One aspect of the present invention is a communication system, which includes a robot installed on a user side and a server, and the robot collects a microphone that collects a user's voice and a user who collects sound using the microphone. The voice signal of the user is transmitted to the server through the network, and the user's voice signal transmitted through the network is transmitted and received as the utterance of the robot. A voice editing unit that edits and generates an edited voice signal; and at least one speaker that outputs the edited voice signal; and the server includes a connection management unit that manages transmission and reception of signals of the robot. It has a communication system.

  One aspect of the present invention is a conversation log collection system in a communication system, which includes a robot installed on a user side and a server, and the robot includes a microphone that collects a user's voice and the microphone. The voice signal of the collected user's voice is transmitted to the server through a network, and the voice signal of the user sent through the network is transmitted / received to receive the user's voice signal sent through the network, The server includes voice editing means that edits the speech of the robot and generates an edited voice signal, and at least one speaker that outputs the edited voice signal, and the server manages transmission and reception of the robot signal. A connection management means that recognizes the user's voice signal sent through the network A speech recognition means for the character string voice is text of the user, a conversation log collecting system and a conversation log collecting means for collecting a conversation log.

  One aspect of the present invention is a conversation log collection server, wherein a microphone that collects a user's voice and a voice signal of the user's voice collected by the microphone are transmitted to the server through a network, and the network Transmitting / receiving means for receiving a user's voice signal sent through the network; voice editing means for editing the user's voice signal sent through the network as an utterance of the robot and generating an edited voice signal; and the editing Connection management means for managing transmission / reception of signals between robots having at least one speaker for outputting voice signals, and voice recognition means for voice recognition of user voice signals sent through a network and converting them into text And a conversation log collecting means for collecting, as a conversation log, a character string in which the user's voice is converted into text. It is a story log collection server.

  One embodiment of the present invention is a communication method, in which a robot installed on a uttering user side collects a user's voice using a microphone, and the robot installed on the uttering user side uses the microphone. The collected voice signal of the user's voice is transmitted to the server via the network, and the server transmits the received voice signal of the user's voice to the destination robot, and the destination robot Receives a voice signal of a user sent through the network, and the destination robot edits the voice signal of the user as an utterance of the robot, generates an edited voice signal, and The talking robot is a communication method for outputting the edited voice signal.

  The present invention can improve the quality of a conversation performed between a user and a robot.

FIG. 1 is a diagram schematically showing a communication robot system according to a first embodiment of the present invention. FIG. 2 is a block diagram showing the configuration of the robot 1. FIG. 3 is a block diagram of the server 3 according to the first embodiment. FIG. 4 is a diagram showing an example of the connection management database 32. FIG. 5 is a diagram showing a modification of the server 3. FIG. 6 is a diagram illustrating an example of the connection management database 32 according to the modification of the first embodiment. FIG. 7 is a block diagram of the server 3 in the second embodiment. FIG. 8 is a block diagram of the server 3 according to the third embodiment.

<First Embodiment>
A first embodiment of the present invention will be described.

  FIG. 1 is a diagram schematically showing a communication robot system according to a first embodiment of the present invention.

  In FIG. 1, 1 is a robot installed on the user A side, 2 is a robot installed on the user B side, and 3 is a server to which the robot 1 and the robot 2 are connected.

  Since the robot 1 and the robot 2 are the same, the configuration of the robot will be described using the robot 1 as an example.

  FIG. 2 is a block diagram showing the configuration of the robot 1.

  As shown in FIG. 2, the robot 1 includes a microphone 11, a voice editing unit 12, a speaker 13, and a control unit 14.

  The microphone 11 is a microphone that collects the voice of the user A.

  The voice editing unit 12 edits the voice signal of the user B transmitted from the server 3 through the network as an utterance of the robot 1 and generates an edited voice signal. Here, editing the user B's voice signal as the utterance of the robot 1 means that the user B's voice (timbre or voice color) is edited (converted) into the robot 1 voice (tone or voice color) with respect to the user B voice signal. ) For example, the voice of a male or female user is edited (converted) into neutral voice unique to the robot, or edited (converted) into the voice of the robot customized by the user.

  The speaker 13 is at least one speaker that outputs an edited audio signal edited (converted) by the audio editing unit 12.

  FIG. 3 is a block diagram of the server 3 according to the first embodiment.

  The server 3 includes an inter-robot connection management unit 31 and a connection management database 32.

  As shown in FIG. 4, the connection management database 32 stores robot identification information (ID), connection status (connected or disconnected), and connected robot identification information (ID) in association with each other.

  The inter-robot connection management unit 31 uses the connection management database 32 to manage the connection between the robots, in this example, the robot 1 and the robot 2.

  Next, the operation of the present embodiment will be described. Note that the communication connection between the robot 1 and the robot 2 is managed by the inter-robot connection management unit 31 of the server 3.

  First, the user A speaks toward the robot 1. The voice of the user A is collected by the microphone 11 and the voice signal is transmitted to the server 3. For example, when the user A talks to “I will make and eat an apple pie”, a voice signal “I will make and eat an apple pie” is transmitted to the server 3.

  The inter-robot connection management unit 31 of the server 3 refers to the connection management database 32 and transmits the audio signal transmitted from the robot 1 to the robot 2.

  In the robot 2, the voice editing unit 12 edits the received voice signal as an utterance of the robot 2 to generate an edited voice signal. For example, the received voice signal “I will make and eat an apple pie” is edited as the utterance of the robot 2 to generate an edited voice signal. Then, the edited audio signal is output from the speaker 13. For example, “I will make and eat an apple pie” is output with the voice specific to the robot 2.

  Here, the user B responds to the voice uttered by the robot 2 and speaks toward the robot 2. The voice of the user B is collected by the microphone 11 and the voice signal is transmitted to the server 3. For example, when the user B replies “I want to eat” to “I will make and eat an apple pie”, a voice signal “I want to eat” is transmitted to the server 3. .

  The inter-robot connection management unit 31 of the server 3 refers to the connection management database 32 and transmits the audio signal transmitted from the robot 2 to the robot 1.

  In the robot 1, the voice editing unit 12 edits the received voice signal as an utterance of the robot 1 to generate an edited voice signal. For example, the received speech signal “Like. I want to eat” is edited as the utterance of the robot 1 to generate an edited speech signal. Then, the edited audio signal is output from the speaker 13. For example, “Like, I want to eat” is output with the voice unique to the robot 1.

  Next, the user A responds to the voice uttered by the robot 1 and speaks toward the robot 1. The voice of the user A is collected by the microphone 11 and the voice signal is transmitted to the server 3. For example, when the user A responds to “Like. I want to eat.”, A voice signal “A. I forgot to buy puff pastry.” Is sent to the server 3. Is done.

  The inter-robot connection management unit 31 of the server 3 refers to the connection management database 32 and transmits the audio signal transmitted from the robot 1 to the robot 2.

  In the robot 2, the voice editing unit 12 edits the received voice signal as an utterance of the robot 2 to generate an edited voice signal. For example, the received audio signal “A. I forgot to buy pie dough” is edited as the utterance of the robot 2 to generate an edited audio signal. Then, the edited audio signal is output from the speaker 13. For example, “Ah, I forgot to buy puff pastry” is output by a voice specific to the robot 2.

  Subsequently, the user B responds to the voice uttered by the robot 2 and speaks toward the robot 2. The voice of the user B is collected by the microphone 11 and the voice signal is transmitted to the server 3. For example, when the user B responds to “Ah, I forgot to buy the pie dough”, the voice signal “Sorry. This time” is transmitted to the server 3.

  Finally, in the robot 1, the voice editing unit 12 edits the received voice signal as an utterance of the robot 1 to generate an edited voice signal. For example, the received voice signal of “sorry. This time” is edited as the utterance of the robot 1 to generate an edited voice signal. Then, the edited audio signal is output from the speaker 13. For example, “unfortunate.

  Thus, compared to engines using a dictionary type, a log type, a Markov sentence generation type, etc., since humans are talking through a robot, natural conversation can be performed. Moreover, since the person with whom a person talks directly is a robot, the conversation can be performed in an atmosphere different from the conversation between persons.

  Next, a modification of the above-described first embodiment will be described.

  FIG. 5 is a diagram showing a modification of the server 3.

  The modification of the first embodiment includes a matching control unit 33 in addition to the inter-robot connection management unit 31 of the server 3 in the first embodiment.

  In addition to the contents of the first embodiment, the connection management database 32 stores attribute information of the user who is the owner of the robot. FIG. 6 shows an example of the connection management database 32 in the modification of the first embodiment. Here, the user attribute information is the age, sex, address, hobby, etc. of the user who owns the robot.

  Based on the user attribute information in the connection management database 32, the matching control unit 33 performs matching between users who wish to have a conversation via the robot and establishes a connection between the robots. For example, when the user A is a woman in her 20s and wants to have a conversation with a similar female user, the matching control unit 33 refers to the user attribute information in the connection management database 32, is a woman in her 20s, and Search the identification information of the robot being cut and attempt to establish a connection with the corresponding robot. If the connection is successfully established, the above-described conversation operation is started. If connection establishment is unsuccessful, another robot is searched and connection establishment is attempted.

  By adopting such a configuration, it is possible to fulfill the wishes of the conversation partner of the robot user.

  Further, the server 3 may include the voice editing unit 12 included in the robot 1. In this case, the server 3 edits the received voice signal of the user for the robot 2 and transmits it to the robot 2. Then, the robot 2 outputs the edited audio signal from the speaker 13.

  Even if comprised in this way, the effect similar to 1st Embodiment mentioned above can be acquired.

<Second Embodiment>
A second embodiment of the present invention will be described.

  FIG. 7 is a block diagram of the server 3 in the second embodiment.

  The server 3 includes a speech recognition unit 34, a filtering unit 35, and an NG word database 36 in addition to the first embodiment.

  The NG word database 36 is a database in which a group of inappropriate terms (hereinafter referred to as prohibited terms) is stored in a conversation.

  The voice recognition unit 34 converts the voice signal transmitted from the robot 1 into text using a conventional voice recognition technique.

  When the prohibited word is included in the text string of the user's voice, the filtering unit 35 deletes the corresponding voice of the user or converts the corresponding voice to another term. Then, the deleted and converted voice is transferred to the communication destination robot.

  In this way, even if inappropriate words are uttered during the conversation, the conversation partner is not made uncomfortable.

<Third Embodiment>
A third embodiment of the present invention will be described.

  A characteristic point of the third embodiment is that it is possible to collect basic data of a conversation engine using a dictionary type (scenario type), a log type, a Markov sentence generation type (text generation type), etc. It is. In particular, in the case of the log type, it is indispensable to collect a large number of conversation logs in order to establish a natural conversation between a human and a robot. Therefore, the third embodiment is characterized in that conversation logs obtained during the conversation are collected using the communication system in the first embodiment.

  FIG. 8 is a block diagram of the server 3 according to the third embodiment.

  The server 3 in the third embodiment includes a conversation log collection unit 37 and a conversation log database 38 in addition to the first and second embodiments.

  The conversation log collection unit 37 collects conversations between the robots 1 and 2 converted into text in the voice recognition unit 34 in the conversation log database 38 as a conversation log.

  Thereby, conversation logs of many natural exchanges can be collected. The collected conversation log can be used for data such as a log-type conversation engine.

  In the above-described embodiment, each unit is configured by hardware, but may be configured by a program that causes the information processing apparatus (CPU) to perform the above-described operation processing.

  Although the present invention has been described with reference to the preferred embodiments, the present invention is not necessarily limited to the above-described embodiments, and various modifications can be made within the scope of the technical idea.

DESCRIPTION OF SYMBOLS 1 Robot 2 Robot 3 Server 11 Microphone 12 Voice editing part 13 Speaker 14 Control part 31 Inter-robot connection management part 32 Connection management database 33 Matching control part 34 Voice recognition part 35 Filtering part 36 NG word database 37 Conversation log collection part 38 Conversation log The database

Claims (10)

A communication system,
A robot installed on the user side and a server;
The robot is
A microphone that collects the user's voice;
A transmission / reception means for transmitting a voice signal of a user's voice collected by the microphone to the server through a network, and receiving a user's voice signal transmitted through the network;
A voice editing means for editing a user's voice signal sent through the network as an utterance of the robot and generating an edited voice signal;
At least one speaker for outputting the edited audio signal;
Have
The server
A communication system comprising connection management means for managing transmission and reception of signals of the robot. The server
A user attribute information database storing attribute information of a user who owns the robot;
The communication system according to claim 1, further comprising: a matching control unit that refers to the user attribute information database, performs user matching based on user attribute information, and establishes a connection between robots corresponding to the matched users. The server
A banned term database that contains banned terms for conversations;
Voice recognition means for recognizing and converting a user's voice signal sent through the network into text;
Filtering means for referring to the prohibited term database, searching for a prohibited term from a character string in which the user's voice is converted into text, and deleting the user's voice signal or converting a part thereof when the prohibited term is included. The communication system according to claim 1 or 2, comprising: The server
Voice recognition means for recognizing and converting a user's voice signal sent through the network into text;
The communication system according to claim 1, further comprising: a conversation log collecting unit that collects a character string in which the user's voice is converted into a text as a conversation log. A conversation log collection system in a communication system,
A robot installed on the user side and a server;
The robot is
A microphone that collects the user's voice;
A transmission / reception means for transmitting a voice signal of a user's voice collected by the microphone to the server through a network, and receiving a user's voice signal transmitted through the network;
A voice editing means for editing a user's voice signal sent through the network as an utterance of the robot and generating an edited voice signal;
At least one speaker for outputting the edited audio signal;
Have
The server
Connection management means for managing transmission and reception of signals of the robot;
Voice recognition means for recognizing and converting a user's voice signal sent through the network into text;
A conversation log collection system comprising conversation log collection means for collecting, as a conversation log, a character string in which the user's voice is converted into text. A conversation log collection server,
A microphone for collecting user's voice; and a transmission / reception means for transmitting a voice signal of the user's voice collected by the microphone to the server through a network and receiving the user's voice signal sent through the network A robot having voice editing means for editing a voice signal of a user sent through a network as an utterance of the robot and generating an edited voice signal; and at least one speaker for outputting the edited voice signal Connection management means for managing transmission and reception of signals between
Voice recognition means for recognizing and converting a user's voice signal sent through the network into text;
A conversation log collection server having a conversation log collection means for collecting a character string in which the user's voice is converted into a text as a conversation log. A communication method,
The robot installed on the uttering user's side collects the user's voice with a microphone,
The robot installed on the side of the user who speaks transmits the voice signal of the user's voice collected by the microphone to the server through the network,
The server transmits an audio signal of the received user's voice to the destination robot,
The destination robot receives the user's voice signal sent through the network,
The destination robot edits the user's voice signal as the robot's utterance to generate an edited voice signal,
A communication method in which the destination robot outputs the edited audio signal. The server
The user attribute information database storing the attribute information of the user who owns the robot is referred to, the user is matched based on the attribute information of the user, and the connection is established between the robots corresponding to the matched user. The communication method described. The server
Voice recognition of user's voice signal sent through the network is recognized and converted into text.
Refer to a prohibited term database in which prohibited terms of conversation are stored, and search for prohibited terms from a text string in which the user's voice is converted to text. The communication method according to claim 7 or 8, wherein the part is converted. The server
Voice recognition of user's voice signal sent through the network is recognized and converted into text.
The communication method according to claim 7, wherein character strings in which the user's voice is converted into text are collected as a conversation log.