JP2019022131A - Speech voice collection method for receiving speech voices for each group, system, speech analysis server and program - Google Patents

Abstract

To provide a speech voice collection method or the like capable of receiving speech voices for each group in such a manner that analysis of speech voices in a group work is not congested.SOLUTION: Multiple groups consist of multiple users. A speech voice collection method includes: a first step of each user terminal transmitting to a speech analysis server voice interval detection information making a terminal identifier and a group identifier correspondent to one or more speech intervals for each continuous predetermined voice interval; a second step of the speech analysis server receiving multiple pieces of speech interval detection information from multiple user terminals for each group and determining a speech voice reception order of the groups based on a predetermined condition in multiple speech intervals included in a voice interval; and a third step of the speech analysis server transmitting a speech voice request corresponding to the speech interval to the multiple user terminals for the unit of groups in the speech voice reception order and receiving speech voices from the user terminals.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a technique for analyzing user utterances in group work.

In recent years, critical thinking skills and problem solving skills have become important as 21st century skills in the fields of education and business. For this purpose, attention is focused on a group work format based on active learning, not a seminar format that provides knowledge unilaterally.
The group work format is based on collaborative learning in which, for example, 3 to 4 people form one group and exchange ideas while continuing creative discussions toward solving problems. Here, the teacher who manages the group work needs to propose effective advice to the group in which active discussion is not conducted.

On the other hand, in the case of the group work format, one teacher cannot observe the learning situation of discussion of all groups. For this purpose, there is a demand for a technique for simultaneously grasping the learning status of a plurality of groups using ICT (Information and Communication Technology).
For example, there is a technique in which an utterance analysis server collects utterance voices of each user in all groups, texts the utterance voices by voice recognition processing, and provides the analysis results to the teacher by utterance analysis processing. The teacher can observe the learning status while looking at the analysis results for each group.

  FIG. 1 is a system configuration diagram in group work.

  According to FIG. 1, a plurality of groups to be discussed by 3 to 4 users are configured. According to the system of FIG. 1, the utterance analysis server 1 is connected to a plurality of user terminals 2 and an administrator terminal 3 via an access point 4.

The utterance analysis server 1 receives the utterance voice from the user terminal 2 and analyzes the utterance voice for each group.
The user terminal 2 is a microphone device that is held or worn by each user, and specifically, is a smartphone that records the voice of the user.
The administrator terminal 3 acquires the analysis result for each group from the utterance analysis server 1 and clearly indicates it to the teacher.
One access point 4 is connected to a WAN (Wide Area Network), the other is connected to a LAN (Local Area Network), and accommodates a plurality of user terminals 2 via the LAN. The network between the access point 4 and the user terminal 2 may be a wireless / wired LAN, a Bluetooth (registered trademark), or a carrier access network.

  The utterance analysis server 1 requires a certain amount of time for speech recognition and utterance analysis processing. For this reason, when the user terminal 2 randomly uploads speech to the speech analysis server 1, a delay due to a temporary increase in speech recognition processing or speech analysis processing due to a lack of speech for each group There will be a delay due to waiting.

  Conventionally, there is a technique for reducing the frequency of processing delay due to an increase in service waiting time for congestion in a large number of mobile terminals (see, for example, Patent Document 1). According to this technique, a receiving device is provided between the terminal and the service providing device. The accepting device accepts the terminal identification number in a queue and notifies the service providing device of the queue. The service providing apparatus provides a service based on the identification number so that the mobile terminal is a slave so that a congestion state does not occur.

JP 2007-133912 A

In general, according to the user terminal 2, the timing of uploading the speech voice is not controlled. Therefore, when uttered voices are simultaneously transmitted from a large number of user terminals 2, network congestion due to increased traffic, congestion due to increased speech recognition processing or waiting delay in utterance analysis processing may occur, and the connection may be blocked. is there.
In particular, when group work is performed by a large number of users at the same time in the same room, a large number of user terminals 2 are connected under the access point 4. In this case, the transmission capacity of the wireless LAN is insufficient and speech data from the user terminal 2 to the speech analysis server 1 is concentrated on the access point 4, and as a result, congestion is likely to occur.

  In addition, the teacher who manages the group work observes the group so that the discussion is not active while looking at the analysis result for each group. Therefore, as much as possible, the teacher wants to provide advice according to the analysis results in order from the group with less active discussion.

  Therefore, the present invention provides an utterance voice collection method, system, utterance analysis server, and program for preferentially receiving utterance voice from a group where discussion is not active while preventing congestion in the analysis of the utterance voice in group work. The purpose is to provide.

According to the present invention, in a utterance voice collection method of a system having a plurality of user terminals that record utterance voices for each user and an utterance analysis server that receives utterance voices from each user terminal via a network,
There are multiple groups of multiple users,
A first step in which each user terminal transmits voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments to a speech analysis server for each predetermined continuous speech segment;
A speech analysis server receives a plurality of speech segment detection information from a plurality of user terminals for each group, and determines a speech speech reception order of groups based on predetermined conditions in a plurality of speech segments included in the speech segment And the steps
The utterance analysis server includes a third step of transmitting an utterance voice request corresponding to an utterance section to a plurality of user terminals in units of groups of the utterance voice reception order, and receiving the utterance voice from each user terminal. And

According to another embodiment of the speech voice collection method of the system of the present invention,
Regarding the second step, it is preferable that the speech reception order of the groups based on the predetermined condition is the order in which the total time or the average time of the speech sections of a plurality of users in each group is short.

According to another embodiment of the speech voice collection method of the system of the present invention,
Create a probability distribution based on the total time or average time of the utterance sections of a plurality of users in a number of past groups in advance,
Regarding the second step, the speech reception order of the groups based on the predetermined condition is the order in which the total time or average time of the speech sections of the plurality of users in each group is out of the predetermined probability range of the probability distribution. Is also preferable.

According to another embodiment of the speech voice collection method of the system of the present invention,
Regarding the second step, the speech reception order of the groups based on the predetermined condition may be the order in which the “number of long-time speech sections” that are equal to or longer than the predetermined threshold time is small for the speech sections of a plurality of users in each group. preferable.

According to another embodiment of the speech voice collection method of the system of the present invention,
Regarding the second step, the speech reception order of the group based on the predetermined condition is equal to or greater than a predetermined threshold time for the “number of short-time speech sections” shorter than the predetermined threshold time for the speech sections of a plurality of users in each group. It is also preferable that the ratio of “the number of long utterance sections” is in ascending order.

According to another embodiment of the speech voice collection method of the system of the present invention,
As for the second step, the speech reception order of the group based on the predetermined condition is calculated as an absolute difference time that is a difference between the total time or average time of the utterance section of each user and the predetermined threshold time, and the total time is large. It is also preferred that the order be.

According to another embodiment of the speech voice collection method of the system of the present invention,
Regarding the second step, the speech reception order of the group based on the predetermined condition is as follows:
Create a probability distribution based on the total time or average time of the utterance sections of a plurality of users in a number of past groups in advance,
It is also preferable that the total value of the probability values in the probability distribution is calculated for each user in the group, and the total value is in ascending order.

According to another embodiment of the speech voice collection method of the system of the present invention,
The utterance section is also preferably the start time and end time of the user's utterance.

According to the present invention, a plurality of user terminals that record speech for each user;
An utterance analysis server that receives uttered voices from each user terminal via a network;
An administrator terminal that manages multiple groups,
An utterance voice collection method for a system having a notification server that transmits an utterance voice request corresponding to an utterance section to a plurality of user terminals,
There are multiple groups of multiple users,
A first step in which each user terminal transmits voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments to a speech analysis server for each predetermined continuous speech segment;
The utterance analysis server receives a plurality of speech section detection information from a plurality of user terminals for each group, determines a group speech speech reception order based on a predetermined condition in a plurality of utterance sections included in the speech section, and A second step of transmitting the utterance voice reception order to the administrator terminal;
A third step in which the administrator terminal transmits a group identifier based on the utterance voice reception order to the notification server in response to an operation of the administrator;
A fourth step in which the notification server transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals belonging to the group identifier received from the administrator terminal;
Each user terminal that has received the utterance voice request has a fifth step of transmitting the utterance voice to the utterance analysis server.

According to the present invention, in a system having a plurality of user terminals that record utterance voices for each user, and an utterance analysis server that receives utterance voices from each user terminal via a network,
There are multiple groups of multiple users,
Each user terminal
Voice segment detection information transmitting means for transmitting to the speech analysis server voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments for each predetermined continuous voice segment;
An utterance voice transmitting means for transmitting an utterance voice to the utterance analysis server when an utterance voice request is received from the utterance analysis server;
Utterance analysis server
Utterance voice reception order determination means for receiving a plurality of voice section detection information from a plurality of user terminals for each group and determining a utterance voice reception order of the group based on a predetermined condition in a plurality of utterance sections included in the voice section; ,
It is characterized by having speech voice receiving means for transmitting a speech voice request corresponding to a speech section to a plurality of user terminals and receiving a speech voice from each user terminal for each group of speech voice reception order.

According to the present invention,
Multiple user terminals that record speech for each user;
An utterance analysis server that receives uttered voices from each user terminal via a network;
An administrator terminal that manages multiple groups,
A system having a notification server that transmits an utterance voice request corresponding to an utterance section to a plurality of user terminals,
There are multiple groups of multiple users,
Each user terminal
Voice segment detection information transmitting means for transmitting to the speech analysis server voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments for each predetermined continuous voice segment;
An utterance voice transmitting means for transmitting the utterance voice to the utterance analysis server when the utterance voice request is received from the notification server;
Utterance analysis server
Utterance voice reception order determination means for receiving a plurality of voice section detection information from a plurality of user terminals for each group and determining a utterance voice reception order of the group based on a predetermined condition in a plurality of utterance sections included in the voice section; ,
An utterance voice reception order transmission means for transmitting the utterance voice reception order to the administrator terminal;
Utterance voice receiving means for receiving utterance voice from the user terminal,
In response to the operation of the administrator, the administrator terminal sends a group identifier based on the speech reception order to the notification server,
The notification server transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals belonging to the group identifier received from the administrator terminal.

According to the present invention, in a utterance analysis server that receives uttered voices from a plurality of user terminals that record uttered voices for each user via a network,
There are multiple groups of multiple users,
Voice segment detection information receiving means for receiving, from each user terminal, voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments for each predetermined continuous voice segment;
Utterance voice reception order determination means for determining a utterance voice reception order of a group based on a predetermined condition in a plurality of utterance sections included in the voice section, using a plurality of voice section detection information;
It is characterized by having speech voice receiving means for transmitting a speech voice request corresponding to a speech section to a plurality of user terminals and receiving a speech voice from each user terminal for each group of speech voice reception order.

According to the present invention, in a program for functioning a computer mounted on an utterance analysis server that receives uttered speech from a plurality of user terminals that record uttered speech for each user via a network,
There are multiple groups of multiple users,
Voice segment detection information receiving means for receiving, from each user terminal, voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments for each predetermined continuous voice segment;
Utterance voice reception order determination means for determining a utterance voice reception order of a group based on a predetermined condition in a plurality of utterance sections included in the voice section, using a plurality of voice section detection information;
The computer is caused to function as an utterance voice receiving unit that transmits an utterance voice request corresponding to an utterance section to a plurality of user terminals in groups of the utterance voice reception order, and receives the utterance voice from each user terminal.

  According to the utterance voice collection method, system, utterance analysis server, and program of the present invention, congestion is not generated in the utterance voice analysis in the group work, and the utterance voice is preferentially received from the group where the discussion is not active. be able to.

It is a system configuration figure in group work. It is a 1st sequence diagram in this invention. It is explanatory drawing showing the audio | voice area which divided the group work time. It is explanatory drawing showing the speech area for every user which divided the audio | voice area. It is a 2nd sequence diagram in this invention. It is explanatory drawing showing the determination method of the speech reception order of the group in this invention. It is explanatory drawing showing the determination method of the speech audio | voice reception order based on the speech pattern of the user in the group in this invention. It is a function block diagram of the speech analysis server in this invention, and a user terminal. It is a system configuration figure containing a notice server in the present invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

As a system assumed by the present invention, a plurality of groups of a plurality of users are configured as in FIG. 1 described above, and the following apparatuses are connected via a network.
Multiple user terminals: Recording of utterance voice for each user Speaking analysis server: receiving utterance voice from each user terminal and utterance analysis for each group According to the present invention, there is congestion in uploading utterance voice and server processing in group work. In order not to occur, the utterance analysis server uploads the utterance voices in order from the user terminals belonging to the group in which discussion within the group is assumed not to be active. The utterance analysis server 1 executes speech recognition processing and utterance analysis processing on the uttered speech in order from such a group, and clarifies the analysis result to the administrator terminal operated by the teacher.

  FIG. 2 is a first sequence diagram in the present invention.

(S1) Each user terminal 2 belongs to one of the groups, and detects one or more utterance sections for each predetermined continuous voice section from the utterance voice of each user during the group work time.

FIG. 3 is an explanatory diagram showing a voice section in which group work time is divided.
According to FIG. 3, for example, a group work time of 30 minutes is divided into, for example, a voice interval of 5 minutes. When the group work is started, the user terminal 2 can record a speech voice. In addition, each user terminal repeats the start and end of recording of uttered voice for each voice section.

FIG. 4 is an explanatory diagram showing an utterance section for each user into which voice sections are divided.
According to FIG. 4, the voice section for each group includes the speech sections of a plurality of users. The utterance section is represented by “start time and end time” at which voice is detected as a user's utterance.

For example, the user terminal 2 sets the utterance start time when the volume exceeds a predetermined threshold, and sets the utterance end time when the volume lower than the predetermined threshold continues for a predetermined time or more. In this way, the utterance section is divided. When one user utters a plurality of times within one voice section, voice detection information including the utterance section for each utterance is generated.
Here, the speech detection method based on sound volume (total power) has been described as an embodiment. However, as other embodiments, various existing technologies such as those based on low-frequency band power, frequency spectrum, and zero crossing can be used. Can be applied.

The voice section detection information is referred to as “VAD (Voice activity detection) information”. The VAD information is, for example, short text data including the following information.
“Group ID” “User (terminal) ID” “Speaking section”

  All user terminals 2 transmit VAD information to the utterance analysis server 1 every time one voice section ends. Since the VAD information is only short text data, it can be transmitted to the speech analysis server 1 in a very short time.

(S2) The utterance analysis server 1 receives a plurality of VAD information from a plurality of user terminals. The utterance analysis server 1 refers to the group number and classifies the VAD information for each group.
Then, the utterance analysis server 1 determines the “utterance voice reception order” of the group based on a predetermined condition in a plurality of utterance sections included in the voice section.
The speech reception order is determined in order from the group in which discussion is not active, and will be described later with reference to FIG.
According to FIG. 2, it is assumed that the “utterance voice reception order” is determined in the order of groups 2->3-> 1. Here, it is determined that the discussion in group 2 is the least active.

(S3) The utterance analysis server 1 transmits an utterance voice request to the plurality of user terminals 2 in units of groups in the utterance voice reception order. According to FIG. 2, first, an utterance voice request is transmitted to a plurality of user terminals 2 belonging to group 2.
On the other hand, each user terminal 2 of the group 2 that has received the utterance voice request transmits the utterance voice to the utterance analysis server 1.
The utterance analysis server 1 analyzes utterances by speech recognition processing and utterance analysis processing for the group 2 that has received the utterance speech.
And the utterance analysis server 1 will transmit an utterance audio | voice request | requirement to the several user terminal 2 of the following group 3 according to an utterance audio | voice reception order, if the analysis of the group 2 is complete | finished.
As described above, the utterance analysis server 1 uploads the utterance voices of the voice sections recorded by the user terminals 2 in the order of the group unit corresponding to the utterance voice reception order.

  According to the present invention, since the uttered voice is uploaded to the utterance analysis server 1 via the access point in units of groups (that is, all user terminals in the group), congestion does not occur in the LAN. Note that the utterance voice in units of groups may be uploaded by one user terminal 2 or may be simultaneously uploaded to a plurality of user terminals 2 in units of groups.

  FIG. 5 is an explanatory diagram showing a method for determining the speech reception order of groups in the present invention.

  The speech reception order is based on a determination index for activation of discussion within the group. In the following, an embodiment of a method for determining the order of receiving seven speech voices will be described.

<First Embodiment: Order Determination Method Based on Total Time of Speech Section>
According to Fig.5 (a), a speech audio | voice reception order is an order with the short total time of the speech area of the some user in each group.
Even for the same speech section, the shorter the total time of the speech sections in the group, the shorter the processing time for speech recognition and speech analysis.
According to the first embodiment, since discussion within the group is not active, the speech time can be collected and analyzed earlier in the group in which the total duration of the speech period is shorter than in other groups. This enables the teacher who operates the administrator terminal 3 to quickly know the speech analysis results in order from the group in which discussion is not active.

<Second Embodiment: Order Determination Method Based on Average Time of Speaking Section>
According to FIG.5 (b), a speech audio | voice reception order is an order with short average time of the speech area of the some user in each group.
Average time = total time of utterance section / number of utterances The shorter the average time of the utterance section, the more you talk about your opinion, such as companion (yes), connective words (thus), yes (yes), questions / denies (why) Often not. On the other hand, the longer the average duration of the utterance section, the more often the story has its own opinion, explanation, and content.
According to the second embodiment, since the discussion within the group is not active, a group having a shorter average duration of the utterance section can collect and analyze the utterance voice earlier than the other groups.

<Third Embodiment: Order Determination Method Based on Probability Distribution of Total Time / Average Time of Speech Section>
According to FIG.5 (c), the probability distribution based on the total time or average time of the speech area of the some user in many past groups is produced previously.
And the utterance audio | voice reception order of a group shall be the order which the total time or average time of the utterance area of the some user in each group is separated from the predetermined probability range of probability distribution.
When the probability distribution is a normal distribution, for example, when the deviation from the average of a plurality of groups deviates from ± 1σ, it can be determined that discussion within the group is not active.
According to the third embodiment, since the discussion within the group is not active, the speech voice is collected and analyzed earlier in the group farther from the average probability distribution than in the predetermined probability range than the other groups. Can do.

<Fourth embodiment: Order determination method based on the number of long speech segments>
According to FIG. 5D, the utterance voice reception order of the groups is the order in which “the number of long-time utterance sections” that is equal to or longer than the predetermined threshold time is small for the utterance sections of a plurality of users in each group.
The smaller the number of utterance sections for a long time, the more often they are not speaking their opinions, such as matching, connecting words, approval, questions and denials. On the other hand, as the number of utterance sections for a long time increases, there are many cases where the story has its own opinion, explanation, and content.
According to the fourth embodiment, since the discussion within the group is not active, a group with a smaller number of long-term utterance intervals can collect and analyze utterance speech earlier than other groups.

<Fifth Embodiment: Order Determination Method Based on the Ratio of the Number of Long Speaking Sections>
According to FIG.5 (e), the speech reception order of a group becomes more than predetermined threshold time with respect to "the number of short time speech areas" shorter than predetermined threshold time about the speech area of the some user in each group. The ratio of “the number of long-time utterance sections” is in ascending order.
Proportion = number of long utterance sections / number of short utterance sections The smaller this ratio, the more often they are not speaking their opinions, such as collusion, connecting words, approval, questions / denials, etc. On the other hand, the higher this ratio, the more often people are talking about their opinions, explanations, and contents.
According to the fifth embodiment, since discussions within a group are not active, a group with a smaller ratio can collect and analyze utterance voices earlier than other groups.

  FIG. 6 is an explanatory diagram showing a method for determining the speech reception order based on the speech patterns of the users in the group according to the present invention.

  According to FIG. 6, a group in which the total time or average time of the utterance sections of the users of the group greatly differs among users is determined as the priority speech reception order.

<Sixth Embodiment: Order Determination Method Based on Difference from Predetermined Threshold Time>
According to FIG. 6 (a), the utterance voice reception order of the group calculates an absolute difference time which is a difference between the total time (or average time) of each user's utterance section and a predetermined threshold time, and the total time In descending order. The predetermined threshold time may be an average utterance time of users in the group, or may be an arbitrary time that is assumed.
For example, in the case of FIG. 6A, it is possible to extract a pattern in which the total duration of the utterance section of each user in the group is greatly different.
Further, as a pattern other than FIG. 6A, it is possible to extract the case where the total time of the utterance sections of all the users is shorter than the predetermined threshold time and the total time of the difference is large.
Furthermore, it is possible to extract the case where the total time of all the user's utterance sections is longer than the predetermined threshold time and the total time of the difference is large. In this case, for example, if there is a problem as a discussion, such as a case where the written sentence is read and published for a long time, or a case where the talk is asserted and the talk is not heard for too long Can do.

<Seventh Embodiment: Order Determination Method Based on Difference Using Probability Distribution>
According to FIG.6 (b), the probability distribution based on the total time or average time of the speech area of the some user in many past groups is produced previously. And the utterance voice reception order of the group is the order in which the total value of the probability values in the probability distribution is calculated for each user of the group and the total value is small.
For example, in the case of FIG. 6B, it is possible to extract patterns in which the total time (or average time) of the utterance time of each user in the group is greatly different.
Moreover, as a pattern other than FIG. 6B, it is possible to extract a case where the total time of the utterance sections of all the users in the group is short and the sum of the probability values is small.
On the other hand, it is possible to extract the case where the total time of the utterance sections of all the users in the group is long and the sum of the probability values is small. In this case, for example, if there is a problem as a discussion, such as a case where the written sentence is read and published for a long time, or a case where the talk is asserted and the talk is not heard for too long Can do.

  FIG. 7 is a functional configuration diagram of the utterance analysis server and the user terminal according to the present invention.

Each user terminal 2 includes an utterance voice accumulation unit 20, an audio segment detection information transmission unit 21, and an utterance voice transmission unit 22.
The utterance voice accumulation unit 20 records the utterance voice uttered by the user.
The voice section detection information transmitting unit 21 transmits VAD information (voice section detection information) in which a terminal ID and a group ID are associated with one or more utterance sections to the utterance analysis server for each predetermined continuous voice section. (See S1 in FIG. 2 described above).
When receiving the utterance voice request from the utterance analysis server 1, the utterance voice transmission unit 22 transmits the utterance voice to the utterance analysis server 2 (see S3 in FIG. 2 described above).

The utterance analysis server 1 includes an utterance voice accumulation unit 10, an utterance voice reception order determination unit 11, and an utterance voice reception unit 12.
The utterance voice storage unit 20 stores the utterance voice received from the user terminal 2.
The utterance voice reception order determination unit 11 receives a plurality of VAD information (voice section detection information) from a plurality of user terminals for each group, and utters a group based on a predetermined condition in the plurality of utterance sections included in the voice section. The voice reception order is determined (see S2 in FIG. 2 described above).
The utterance voice receiving unit 12 transmits the utterance voice request corresponding to the utterance section to a plurality of user terminals in units of the utterance voice reception order, and receives the utterance voice from each user terminal (see S3 in FIG. 2 described above). ).

  FIG. 8 is a second sequence diagram in the present invention.

The system of FIG. 8 further includes an administrator terminal 3 and a notification server 5 as compared with FIG.
The manager terminal 3 is operated by a group work teacher and manages a plurality of groups.
The notification server 5 transmits an utterance voice request corresponding to an utterance section to a plurality of user terminals in response to an instruction from the administrator terminal 3.

(S1) Each user terminal transmits VAD information (voice section detection information) in which a terminal identifier and a group identifier are associated with one or more utterance sections to the utterance analysis server 1 for each predetermined continuous voice section. .
(S21) The utterance analysis server 1 receives a plurality of speech segment detection information from a plurality of user terminals for each group, and sets the utterance speech reception order of the groups based on predetermined conditions in the plurality of utterance segments included in the speech segment. decide.
(S22) The utterance analysis server 1 transmits the utterance voice reception order to the administrator terminal 3.
(S23) The administrator terminal 3 transmits a group identifier based on the speech reception order to the notification server 5 according to the operation of the administrator.
(S3) The notification server 5 transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals belonging to the group identifier received from the administrator terminal.
On the other hand, each user terminal that has received the utterance voice request transmits the utterance voice to the utterance analysis server.

Furthermore, the administrator terminal 3 may transmit the utterance voice reception request to the notification server 5 for each terminal ID of the user terminal 2 in accordance with the operation of the administrator.
In this case, the notification server 5 pushes an utterance voice request corresponding to the utterance section to the user terminal having the terminal ID instructed from the administrator terminal 3 in a push manner.
The user terminal 2 that has received the utterance voice request transmits the utterance voice to the utterance analysis server 1.
The utterance analysis server 1 analyzes the utterance voice of the user terminal 2 and transmits the analysis result to the manager terminal 3 together with the terminal ID.
By repeating this, the manager terminal 3 can clearly indicate the utterance analysis result for each user terminal 1 to the manager.

According to the sequence of FIG. 2 described above, the utterance analysis server 1 transmits an utterance voice request to the user terminal 2. However, for example, in the case of HTTP (HyperText Transfer Protocol), a request response type Pull in which a client (browser application installed in a user terminal) transmits a request to a server (utterance analysis server) and returns a response from the server to the client. It becomes a method. Therefore, it is difficult for the utterance analysis server 1 to transmit in a push manner to the user terminal 2 such as a smartphone.
On the other hand, according to the sequence of FIG. 8, compared with the sequence of FIG. 2, the utterance analysis server 1 that receives and analyzes the utterance voice from the user terminal 2, and the notification that transmits the utterance voice request to the user terminal 2 This is shared with the server 5.

  Examples of the push information transmission function in the notification server 5 include JavaScript (registered trademark) Web Notification API, Google (registered trademark) Firebase Notification (registered trademark), and Apple (registered trademark) Push Notification. By installing an application corresponding to these APIs (Application Programming Interface) in the user terminal 2, a server push can be realized.

  FIG. 9 is a system configuration diagram including a notification server in the present invention.

The system of FIG. 9 corresponds to the sequence of FIG.
Each user terminal 2 is different from FIG. 7 only in the utterance voice transmission unit 22. The utterance voice transmitting unit 22 in FIG. 9 transmits the utterance voice to the utterance analysis server 1 when receiving the utterance voice request from the notification server 5.
The speech analysis server 1 is different from the speech analysis server 1 only in the speech reception order transmission unit 13. The utterance reception order transmission unit 13 in FIG. 9 transmits the utterance voice reception order to the administrator terminal 3.

  The administrator terminal 3 receives the utterance reception order from the utterance analysis server 1 and clearly indicates the utterance reception order to the administrator (teacher). Then, according to the operation of the administrator, the group ID based on the speech reception order and the user terminal ID included in the group ID are transmitted to the notification server 5.

  The notification server 5 pushes an utterance voice request to a plurality of user terminals 2 belonging to the group ID received from the administrator terminal 3 in a push manner.

<Other embodiment by the notification server 5>
8 and 9, various controls for the user terminal 2 can be executed using the notification server 5 according to instructions from the administrator terminal 3.
Here, the user terminal 2 has installed a dedicated utterance recording application, and the administrator terminal 3 has also installed a dedicated teacher control application.
The utterance recording application of the user terminal 2 generates a unique terminal ID as an initial setting, and registers the terminal ID and group ID in the utterance analysis server 1 in advance. Then, the utterance recording application can decode the instruction information received from the notification server 5 and activate the function accordingly.
The teacher control application of the administrator terminal 3 clearly indicates the group ID and terminal ID received from the utterance analysis server 1 to the administrator using the display. The teacher control application can transmit instruction information to the user terminal 2 via the notification server 2 for each group ID or each terminal ID.

Specifically, the following other embodiments can be realized.
[First Other Embodiment: Control of Start / End of Utterance Recording for User Terminal 2]
Based on the instruction information from the administrator terminal 3, the notification server 5 synchronizes the start / end of the voice section of the user terminal 2. In order to accurately analyze the utterance of the group work in time series, it is necessary to synchronize the voice sections in the plurality of user terminals 2.
The administrator terminal 3 transmits a voice segment synchronization request (start / end) including all user terminal IDs to the notification server 5 by a button operation for starting / ending utterance recording by the administrator.
In response to this, the notification server 5 transmits a voice segment synchronization request (start / end) to the plurality of user terminals 2 in a broadcast manner.
The user terminal 2 that has received the voice segment synchronization request (start / end) starts or ends the recording of the uttered voice during the group work time.
As a result, the group work time and the voice section can be synchronized with respect to all the user terminals 2, the recording time difference is eliminated, and accurate speech analysis can be performed in time series.

[Second Other Embodiment: Control of Time Width / Separation Timing of Voice Section of User Terminal 2]
Based on the instruction information from the administrator terminal 3, the notification server 5 controls the time width / separation timing of the voice section of the user terminal 2. As a result, the voice section can be variably controlled according to the contents of the group work, not the fixed voice section.
As controllable ones, there are time widths of voice sections (for example, 5 minutes-> 10 minutes) or break timings (for example, 11:05, 11:15, 11:20, etc.).
For example, at first, the duration of the voice section is increased to analyze the utterance situation of the group work, and then the duration of the voice section is reduced to analyze the utterance situation of the group work. As a result, advice can be provided according to the analysis result of the uttered voice as required by the teacher.

[Third Other Embodiment: Recognition of Activation State of User Terminal 2]
Based on the instruction information from the administrator terminal 3, the notification server 5 confirms whether the user terminal 2 is powered on / off.
The administrator terminal 3 transmits an activation confirmation request including all user terminal IDs to the notification server 5 by an activation confirmation button operation by the administrator.
In response to this, the notification server 5 transmits an activation confirmation request to the plurality of user terminals 2 in a broadcast manner.
The user terminal 2 that has received the activation confirmation request transmits an activation response to the utterance analysis server 1.
The utterance analysis server 1 transmits a list of terminal IDs that have received the activation response to the administrator terminal 3.
As a result, the administrator terminal 3 clearly indicates the list of terminal IDs to the administrator, and the administrator can recognize the active terminal ID. This can also be used to check for absentees.

  As explained in detail above, according to the speech voice collection method, system, speech analysis server and program of the present invention, it is possible to prevent congestion in the analysis of speech voice in group work, and from a group where discussion is not active. Speech speech can be received preferentially.

  For the various embodiments of the present invention described above, various changes, modifications, and omissions in the technical idea and scope of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Utterance analysis server 10 Utterance voice storage part 11 Utterance voice reception order determination part 12 Utterance voice reception part 2 User terminal 20 Utterance voice storage part 21 Voice section detection information transmission part 22 Utterance voice transmission part 3 Administrator terminal 4 Access point 5 Notification server

Claims (13)

In a method for collecting speech sounds of a system having a plurality of user terminals that record speech sounds for each user and a speech analysis server that receives speech sounds from each user terminal via a network,
There are multiple groups of multiple users,
A first step in which each user terminal transmits, to the utterance analysis server, voice section detection information in which a terminal identifier and a group identifier are associated with one or more utterance sections for each predetermined continuous voice section;
The utterance analysis server receives a plurality of speech segment detection information from a plurality of user terminals for each group, and determines a group speech speech reception order based on predetermined conditions in a plurality of utterance segments included in the speech segment. Two steps,
The utterance analysis server includes a third step of transmitting an utterance voice request corresponding to the utterance section to a plurality of user terminals and receiving the utterance voice from each user terminal in a group unit of the utterance voice reception order. A speech voice collection method for a system characterized by the above. 2. The second step according to claim 1, wherein the utterance voice reception order of the groups based on the predetermined condition is an order in which the total time or average time of the utterance sections of a plurality of users in each group is short. How to collect voices from the system. Create a probability distribution based on the total time or average time of the utterance sections of a plurality of users in a number of past groups in advance,
Regarding the second step, the speech reception order of the groups based on the predetermined condition is the order in which the total time or average time of the speech sections of a plurality of users in each group is out of the predetermined probability range of the probability distribution. The method of collecting speech of the system according to claim 1, wherein: Regarding the second step, the utterance voice reception order of the groups based on the predetermined condition is the order in which “the number of long-time utterance sections” that is equal to or longer than a predetermined threshold time is small for the utterance sections of a plurality of users in each group. The utterance voice collection method of the system according to claim 1. In the second step, the speech reception order of the groups based on the predetermined condition is equal to or greater than a predetermined threshold time for the “number of short-time speech sections” shorter than the predetermined threshold time for the speech sections of a plurality of users in each group The method according to claim 1, wherein the ratio of “the number of long-term utterance sections” is ascending in order. Regarding the second step, the speech reception order of the group based on the predetermined condition is calculated as an absolute difference time that is a difference between a total time or an average time of each user's utterance section and a predetermined threshold time, and the total time The method according to claim 1, wherein the utterance voice collection method is in the descending order. Regarding the second step, the speech reception order of the group based on the predetermined condition is:
Create a probability distribution based on the total time or average time of the utterance sections of a plurality of users in a number of past groups in advance,
The total value of the probability values in the probability distribution is calculated for each user of the group, and the total value is in ascending order.
The method for collecting speech sounds of the system according to claim 1. The method for collecting utterance voices of a system according to any one of claims 1 to 7, wherein the utterance section is a start time and an end time of a user's utterance. Multiple user terminals that record speech for each user;
An utterance analysis server that receives the uttered voice from each user terminal via a network;
An administrator terminal that manages multiple groups,
An utterance voice collection method of a system having a notification server that transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals,
There are multiple groups of multiple users,
A first step in which each user terminal transmits, to the utterance analysis server, voice section detection information in which a terminal identifier and a group identifier are associated with one or more utterance sections for each predetermined continuous voice section;
The utterance analysis server receives a plurality of voice section detection information from a plurality of user terminals for each group, determines a group voice reception order based on predetermined conditions in a plurality of utterance sections included in the voice section, A second step of transmitting the speech reception order to the manager terminal;
A third step in which the administrator terminal transmits a group identifier based on the speech reception order to the notification server in response to an operation of the administrator;
A fourth step in which the notification server transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals belonging to the group identifier received from the administrator terminal;
A utterance voice collection method for a system, comprising: a fifth step in which each user terminal that has received the utterance voice request transmits the utterance voice to the utterance analysis server. In a system having a plurality of user terminals that record utterance voices for each user and an utterance analysis server that receives the utterance voices from each user terminal via a network,
There are multiple groups of multiple users,
Each user terminal
Voice segment detection information transmitting means for transmitting voice segment detection information in which a terminal identifier and a group identifier are associated with one or more utterance segments to the utterance analysis server for each predetermined continuous speech segment;
Utterance voice transmission means for transmitting the utterance voice to the utterance analysis server when the utterance voice request is received from the utterance analysis server,
The speech analysis server
Utterance voice reception order determination means for receiving a plurality of voice section detection information from a plurality of user terminals for each group and determining a utterance voice reception order of the group based on a predetermined condition in a plurality of utterance sections included in the voice section; ,
A speech voice receiving means for transmitting a speech voice request corresponding to the speech period to a plurality of user terminals and receiving a speech voice from each user terminal in units of groups of the speech voice reception order; . Multiple user terminals that record speech for each user;
An utterance analysis server that receives the uttered voice from each user terminal via a network;
An administrator terminal that manages multiple groups,
A system having a notification server that transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals,
There are multiple groups of multiple users,
Each user terminal
Voice segment detection information transmitting means for transmitting voice segment detection information in which a terminal identifier and a group identifier are associated with one or more utterance segments to the utterance analysis server for each predetermined continuous speech segment;
An utterance voice transmitting means for transmitting an utterance voice to the utterance analysis server when an utterance voice request is received from the notification server;
The speech analysis server
Utterance voice reception order determination means for receiving a plurality of voice section detection information from a plurality of user terminals for each group and determining a utterance voice reception order of the group based on a predetermined condition in a plurality of utterance sections included in the voice section; ,
An utterance voice reception order transmission means for transmitting the utterance voice reception order to the manager terminal;
Utterance voice receiving means for receiving utterance voice from the user terminal,
The manager terminal sends a group identifier based on the utterance voice reception order to the notification server according to an operation of the manager,
The notification server transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals belonging to the group identifier received from the administrator terminal. In the utterance analysis server that receives the uttered voice via a network from a plurality of user terminals that record the uttered voice for each user,
There are multiple groups of multiple users,
Voice segment detection information receiving means for receiving, from each user terminal, voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments for each predetermined continuous voice segment;
Utterance voice reception order determination means for determining a utterance voice reception order of a group based on a predetermined condition in a plurality of utterance sections included in the voice section using the plurality of voice section detection information;
Utterance voice receiving means for transmitting an utterance voice request corresponding to the utterance section to a plurality of user terminals and receiving the utterance voice from each user terminal in a group unit of the utterance voice reception order Analysis server. In a program for functioning a computer installed in an utterance analysis server that receives the utterance voice via a network from a plurality of user terminals that record utterance voice for each user,
There are multiple groups of multiple users,
Voice segment detection information receiving means for receiving, from each user terminal, voice segment detection information in which a terminal identifier and a group identifier are associated with one or more speech segments for each predetermined continuous voice segment;
Utterance voice reception order determination means for determining a utterance voice reception order of a group based on a predetermined condition in a plurality of utterance sections included in the voice section using the plurality of voice section detection information;
The computer is caused to function as an utterance voice receiving unit that transmits an utterance voice request corresponding to the utterance section to a plurality of user terminals in units of groups of the utterance voice reception order, and receives the utterance voice from each user terminal. A program for the utterance analysis server.

Images