JP2016218522A - Schedule preparation device, schedule preparation method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prepare an exact shared schedule.SOLUTION: A schedule preparation device 200 comprises: an extraction unit 22 for extracting a keyword belonging to a category related to a schedule from conversation voice of a plurality of speakers; an estimation unit 23 for, on the basis of whether the extraction unit 22 extracts a plurality of different keywords belonging to the same category, estimating the presence or absence of schedule change in the conversation voice; a selection unit 24 for, in the case where the estimation unit 23 estimates the presence of schedule change, selecting a keyword to be used for a shared schedule shared among the plurality of speakers, out of the plurality of different keywords belonging to the same category; and a schedule preparation unit 25 for preparing a shared schedule on the basis of the keyword selected by the selection unit 24.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a schedule creation device, a schedule creation method, and a program.

Conventionally, schedule management using a PC (Personal Computer), a mobile phone, or the like has been widely performed. Recently, a technique for creating a schedule based on voice input by a user is known.
For example, Patent Literature 1 discloses a technique for creating a schedule using words related to a schedule extracted from input speech and additional information of the words.
Patent Document 2 discloses a technology for sharing the contents of voiced text data between users.

JP 2010-198262 A JP 2013-118488 A

By the way, a schedule may be shared by a plurality of related parties. Here, it is conceivable that a schedule shared by a plurality of parties is automatically created by the techniques of Patent Document 1 and Patent Document 2.
However, in a conversation between a plurality of parties, for example, schedule adjustment between the parties may be discussed on the spot. For this reason, compared with the case where a single schedule is created from the voice of one speaker, the accuracy of the sharing schedule is lowered. For this reason, it is desired to create an accurate sharing schedule.

  Therefore, the present invention has been made in view of the above circumstances, and an object thereof is to provide a schedule creation device or the like that creates an accurate sharing schedule.

In order to achieve the above object, a schedule creation device according to the first aspect of the present invention includes:
An extraction means for extracting keywords belonging to a category related to a schedule from conversational voices of a plurality of speakers;
Based on whether the extraction means has extracted a plurality of different keywords belonging to the same category, the estimation means for estimating the presence or absence of a schedule change in the conversation voice;
When the estimating means estimates that there is a schedule change, out of a plurality of different keywords belonging to the same category, a selecting means for selecting a keyword used for a shared schedule shared by the plurality of speakers;
Schedule creation means for creating the shared schedule based on the keyword selected by the selection means;
It is provided with.

In order to achieve the above object, a schedule creation device according to the second aspect of the present invention provides:
Speaker identification means for identifying the speaker;
When the speaker identification means identifies a plurality of speakers, an extraction means for extracting keywords belonging to a category related to the schedule from the conversation voices of the plurality of speakers;
Schedule creation means for creating a sharing schedule that is shared only by the speaker who spoke the keyword extracted by the extraction means;
Schedule correction means for correcting the sharing schedule created by the schedule creation means based on conversational voices of the plurality of speakers;
It is provided with.

  According to the present invention, an accurate sharing schedule can be created.

It is a figure for demonstrating the schedule creation apparatus which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the structure of the user terminal which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the structure of the schedule creation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows an example of the template for speaker identification. It is a figure which shows an example of a keyword table. It is a figure which shows an example of a schedule format. It is a figure which shows an example of the flowchart of the schedule creation process which concerns on the 1st Embodiment of this invention. It is a figure which shows an example of a sharing schedule. It is a block diagram which shows the structure of the schedule preparation apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows an example of the flowchart of the schedule creation process which concerns on the 2nd Embodiment of this invention. It is a figure which shows an example of the single schedule before an update. It is a figure which shows an example of the share schedule after an update. It is a figure for demonstrating the schedule creation apparatus which concerns on the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the schedule preparation apparatus which concerns on the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the schedule preparation apparatus which concerns on the 4th Embodiment of this invention.

(First embodiment)
Hereinafter, an outline of a schedule creation device according to an embodiment of the present invention will be described with reference to FIG. In this embodiment, as shown in FIG. 1, a description will be given on the assumption that the speakers A and B are talking about a schedule at a meeting or the like. In this embodiment, as an example, the schedule creation device 200 will be described as a server, and the user terminal 100 will be described as a smartphone. In the following description, when it is not necessary to specifically identify the speaker A or B in FIG.

  The schedule creation device 200 creates a schedule based on the voice of the speaker. In the case of the scene setting shown in FIG. 1, a schedule is created based on the conversational voices of a plurality of speakers. This voice is obtained by transmitting the voice collected by the user terminal 100 to the schedule creation device 200 in real time.

  Hereinafter, specific configurations of the user terminal 100 and the schedule creation device 200 will be described in order. First, the configuration of the user terminal 100 will be described with reference to FIG.

  As illustrated in FIG. 2, the user terminal 100 includes a control unit 11, an input unit 12, a microphone 13, a camera 14, a storage unit 15, a communication unit 16, and a display unit 17.

  The control unit 11 includes a CPU (Central Processing Unit), and controls the entire user terminal 100 by executing a control program stored in the storage unit 15.

  The input unit 12 includes various operation buttons for inputting user instructions, a touch panel arranged in a superimposed manner on a display included in the display unit 17, a software keyboard displayed on the display, and the like. Examples of the instruction input in this embodiment include a start instruction for instructing the start of the schedule creation process, and an end instruction for instructing the end of the schedule creation process.

  The microphone 13 picks up the voice of the speaker. The camera 14 captures an image of the speaker. The collected voice and the captured image are transmitted to the schedule creation device 200 in real time via the communication unit 16.

  The storage unit 15 includes a RAM (Random Access Memory), a ROM (Read Only Memory), and a nonvolatile memory. The RAM temporarily stores data and programs, and functions as a work memory for the CPU provided in the control unit 11. The ROM stores a control program necessary for controlling the entire user terminal 100. The nonvolatile memory is, for example, a hard disk and stores various data. As various data, for example, the terminal ID (terminal identification information) of the user terminal 100 is included.

  The communication unit 16 transmits / receives data to / from the schedule creation device 200 via an arbitrary communication network (for example, a mobile phone network). Specifically, the communication unit 16 transmits the voice collected by the microphone 13, the image captured by the camera 14, and the terminal ID of the own terminal to the schedule creation device 200.

  The display unit 17 includes a display and displays various images. Examples of the various images include schedules created by the schedule creation device 200.

Next, the configuration of the schedule creation device 200 will be described with reference to FIG.
As illustrated in FIG. 3, the schedule creation device 200 includes a control unit 20, a storage unit 30, and a communication unit 40.

  The control unit 20 includes a CPU, and controls the entire schedule creation device 200 by executing a control program stored in the storage unit 30.

  The storage unit 30 includes a RAM, a ROM, and a nonvolatile memory. The RAM temporarily stores data and programs, and functions as a work memory for the CPU provided in the control unit 20. The ROM stores a control program necessary for controlling the entire schedule creation device 200. The nonvolatile memory is, for example, a hard disk, and stores a schedule creation program and various data. The various data include a speaker identification template 31, a keyword table 32, a schedule format 33, and the like. These will be described later. The storage unit 30 functions as a schedule storage unit 34 that stores the schedule created by the control unit 20.

  The communication unit 40 transmits and receives data to and from the user terminal 100 via an arbitrary communication network (for example, a mobile phone network). Specifically, the communication unit 40 receives the sound collected by the microphone 13, the image captured by the camera 14, and the terminal ID of the user terminal 100 from the user terminal 100.

Next, functions of the control unit 20 will be described.
The control unit 20 functions as a speaker identification unit 21, an extraction unit 22, an estimation unit 23, a selection unit 24, and a schedule creation unit 25 by executing a schedule creation program stored in the storage unit 30.

The speaker identification unit 21 identifies the speaker based on the voice of the speaker, the image of the speaker, and the terminal ID received from the user terminal 100 via the communication unit 40.
Hereinafter, (1) when a speaker is identified by voice recognition, (2) when a speaker is identified by image recognition, and (3) when a speaker is identified by a terminal ID will be described in order.

  First, (1) when a speaker is identified by voice recognition, the speaker identification unit 21 uses any known technique to acquire a feature amount indicating the characteristics of the speaker's voice from the received speaker's voice. . In this embodiment, as an example, a voice print (sound spectrogram) of a speaker is used as a voice feature amount. The speaker identifying unit 21 identifies a speaker based on whether or not the acquired voiceprint of the speaker matches any of the voiceprints of a plurality of registered users learned in advance. The registered user refers to a user who is registered in advance as a user who uses the schedule creation function of the schedule creation device 200.

  Next, (2) when a speaker is identified by image recognition, the speaker identifying unit 21 acquires an image indicating the characteristics of the speaker from the received speaker image using any known technique. In this embodiment, as an example, the face image of the speaker is used as an image indicating the characteristics of the speaker. The speaker identifying unit 21 identifies a speaker based on whether or not the acquired face image of the speaker matches any of the face images of a plurality of registered users learned in advance.

  Next, (3) when the speaker is identified by the terminal ID, the speaker identification unit 21 matches the received terminal ID with the terminal ID associated with the user ID (user identification information) stored in advance. The speaker is identified by whether or not to do so.

  The speaker identification according to the above (1) to (3) is specifically performed using the speaker identification template 31.

  As shown in FIG. 4, the speaker identification template 31 includes a registered user name as a user ID, a voice print of the registered user of the registered user name, a face image of the registered user, and a user used by the registered user. It is the table which matched terminal ID (terminal identification information) of a terminal (user terminal 100 in this embodiment). For example, if the registered user name is “A”, the voiceprint of the registered user A is “A1”, the face image of the registered user A is “A2”, and the terminal ID of the user terminal used by the registered user A is “A3”. I understand that.

The speaker identification unit 21 compares the voice print of the speaker received from the user terminal, the face image of the speaker, and the received terminal ID with the voice print, face image, and terminal ID included in the speaker identification template 31, respectively. To do.
If the speaker identification unit 21 determines that at least one of the voiceprint, face image, and terminal ID matches the voiceprint, face image, or terminal ID included in the speaker identification template 31, the speaker identification unit 21 matches the speaker. The speaker is identified as a speaker with a registered user name associated with a voiceprint, face image, or terminal ID. Then, the speaker identifying unit 21 supplies the registered user name (for example, “A”) of the identified speaker to the schedule creating unit 25.

  Returning to FIG. 3, the extraction unit 22 extracts keywords belonging to the category related to the schedule from the voice of the speaker. When there are a plurality of speakers, the extraction unit 22 extracts the keywords from the conversation voices of the plurality of speakers. Specifically, the extraction unit 22 extracts keywords belonging to a category related to the schedule using the keyword table 32 learned in advance.

  As shown in FIG. 5, the keyword table 32 is a table in which categories related to the schedule are associated with texts of keywords belonging to the categories. In the present embodiment, as shown in FIG. 5, the keyword table 32 includes “month”, “day”, “time”, “location”, and “planned contents” as categories. For example, the category “month” is associated with keywords “January”, “February”, “March”, “April”, and the like.

  The extraction unit 22 converts the voice of the speaker received from the user terminal 100 into text, and compares the converted text with the keyword text in the keyword table 32. When there are a plurality of speakers, the extraction unit 22 converts the conversational voices of the plurality of speakers into text, and the converted text is a text of a keyword belonging to any one of a plurality of categories in the keyword table 32. If they match, the matched text is extracted as a keyword.

  Returning to FIG. 3, the estimation unit 23 estimates whether or not there is a schedule change in the voice (conversation voice if there are a plurality of speakers) based on whether or not the extraction unit 22 has extracted a plurality of different keywords belonging to the same category. To do. The estimation unit 23 estimates the presence / absence of the schedule change by excluding a schedule different from the schedule related to the schedule change. That is, the presence / absence of a schedule change is estimated within the same schedule.

  For example, speaker A in Figure 1 asks, “I would like to have a meeting in the conference room from 3 o'clock on April 1st. If “3 o'clock is more convenient for the meeting”, the meeting, which is the “scheduled content” of the category, matches, so it is estimated that speakers A and B are talking about the same schedule. The estimation unit 23 then changes the schedule for the category “day” in the conversation voices of the speakers A and B by extracting different keywords “1 day” and “2 days” belonging to the category “day” in the same schedule. Presume that there was.

  As described above, the estimation unit 23 estimates whether or not the schedules are the same based on whether or not keywords of a predetermined category (in this embodiment, “planned contents” as an example) match between speakers. The presence or absence of a schedule change within the same schedule is estimated. Note that the criterion for determining whether or not the schedules are the same is an example, and another criterion may be used.

  Next, when the estimation unit 23 estimates that there is a schedule change, the selection unit 24 selects a keyword to be used for schedule creation from among a plurality of different keywords belonging to the same category. In the case of the above-described example, the selection unit 24 selects a keyword to be used for the sharing schedule from different keywords “1 day” and “2 days” belonging to the category “day”. Selection criteria will be described later.

On the other hand, the selection unit 24 selects one keyword when the estimation unit 23 estimates that there is no schedule change, that is, when one keyword is extracted in the same category from conversational voices of a plurality of speakers.
As described above, in this embodiment, the estimation unit 23 and the selection unit 24 improve the accuracy of the schedule (particularly, the shared schedule).

  Next, the schedule creation unit 25 creates a schedule based on the keyword selected by the selection unit 24. The selection unit 24 creates a shared schedule when there are a plurality of speakers and a single schedule when there is only one speaker.

Here, the schedule creation unit 25 creates a schedule by inputting the keyword selected by the selection unit 24 into the schedule format 33.
As shown in FIG. 6, the schedule format 33 is a table in which registered user names and data spaces for storing keywords for each category related to the schedule are associated with each other.

  The schedule creation unit 25 creates the schedule by inputting the keyword selected by the selection unit 24 into the data space associated with the registered user name of the identified speaker. The single schedule is a schedule in which a keyword is input to the data space of one registered user, and the shared schedule is a schedule in which keywords are input in common to the data spaces of a plurality of registered users.

  The schedule creation unit 25 supplies the created schedule to the schedule storage unit 34 of the storage unit 30. In addition, the schedule memorize | stored in the schedule memory | storage part 34 can be referred by accessing from each user terminal which each registered user uses.

  The specific configurations of the user terminal 100 and the schedule creation device 200 have been described above. Hereinafter, the schedule creation process executed by the schedule creation device 200 will be described with reference to FIG. The schedule creation process will be described by taking the scene setting shown in FIG. 1 as an example. Further, the schedule creation device 200 naturally does not know who the speaker is before speaker identification.

First, the speaker A inputs a start instruction for starting schedule creation after starting the schedule creation program from the input unit 12 of the user terminal 100. Then, the user terminal 100 that has received the start instruction starts collecting sound and capturing an image, and transmits the terminal ID of the own terminal together with the start instruction to the schedule creation device 200.
On the other hand, when receiving the start instruction, the schedule creation device 200 enters a voice / image standby state. The schedule creation process is started when a voice / image transmitted from the user terminal 100 is received.

  First, the extraction unit 22 of the schedule creation device 200 extracts keywords (step S11). Specifically, the extraction unit 22 performs the keyword extraction process by the extraction method using the keyword table 32 described above on the voice received in real time.

  Next, the speaker identification part 21 identifies a speaker (step S12). Specifically, the speaker identification unit 21 uses the speaker identification template 31 to perform the above-described speaker identification processing (1) and (2) on the voice / image received in real time. At the same time, the speaker identification unit 21 performs the above-described speaker identification process (3) using the terminal ID and the speaker identification template 31 transmitted together with the start instruction.

  Next, the extraction unit 22 determines whether or not the keyword extraction has ended (step S13). For example, the extraction unit 22 determines whether or not to extract the keyword based on a timeout when the sound is interrupted for a predetermined time or more, or reception of an end instruction transmitted from the user terminal 100. Until this keyword extraction ends (step S13; No), that is, the conversation ends when there are a plurality of speakers, and the processing of steps S11 and S12 is repeated until the utterance ends when the speakers are alone, thereby extracting keywords and talking. Identification of the person is continued (loop of steps S11, S12, S13).

  Here, although steps S11 and S12 are processed in order for convenience of explanation, these extraction and speaker identification are processed simultaneously with respect to voice / image received in real time. In other words, the schedule creation device 200 identifies the speaker by the voice / image / terminal ID while extracting the keyword from the received voice until the keyword extraction is completed.

  Here, as an example 1, in order for speaker A in FIG. 1 to confirm the schedule, he / she hears "Tell me about the time and place for the April 2 meeting." Suppose that it will be held in the meeting room from 3 o'clock.

  In this case, in step S11, the extraction unit 22 selects “April”, “2nd”, “Meeting” that matches the keyword text included in the keyword table 32 from the text converted from the voice received from the user terminal 100. ”,“ 3 o'clock ”, and“ meeting room ”are extracted as keywords related to the schedule.

On the other hand, in step S12, the speaker identification unit 21 uses the speaker identification template 31 to specify the registered user name “A” associated with the terminal ID “A3” transmitted together with the start instruction. Then, the speaker identifying unit 21 identifies that the speaker is a speaker having the registered user name “A”.
In addition, the speaker identification unit 21 determines that the voiceprint obtained from the received voice “Meeting is performed in the conference room from 3 o'clock” matches the voiceprint B1 using the speaker identification template 31; Another speaker is identified as the speaker with the registered user name “B”.

  Returning to the schedule creation process of FIG. 7, when the keyword extraction is completed (step S13; Yes), the speaker identifying unit 21 determines whether there are a plurality of identified speakers (step S14). Here, when it is determined that there are a plurality of identified speakers (step S14; Yes), in order to improve the accuracy of the sharing schedule, the estimation unit 23 and the selection unit 24 cooperate to change the schedule in the conversation voice. A keyword is selected based on the presence or absence (step S15).

  In the case of Example 1 described above, the estimation unit 23 first matches the “meeting” uttered by the speaker A and the “meeting” uttered by the speaker B, and the extracted keyword (“April”, In “2 days”, “meeting”, “3 o'clock”, “meeting room”), there is no plurality of different keywords belonging to the same category, so it is estimated that there is no schedule change. Subsequently, since it is estimated that there is no schedule change, the selection unit 24 selects one keyword (“April”, “2nd day”, “meeting”, “3 o'clock”, “meeting room”) belonging to the same category. Select each one.

Returning to FIG. 7, the schedule creation unit 25 creates a shared schedule based on the selected keyword (step S16). In the case of Example 1 described above, the schedule creation unit 25 determines whether the speakers A and B shown in FIG. 8 are based on “April”, “2 days”, “meeting”, “3 o'clock”, and “meeting room”. Create a sharing schedule.
On the other hand, as example 2, speaker A in Fig. 1 asks, "I would like to have a meeting in the conference room from 3 o'clock on April 1st. “It would be better to have a meeting at 3 o'clock on April 2,” and speaker A replied, “I prefer April 1, but April 2 is fine. ".

  In this case, in step S15, since the “meeting” uttered by the speaker A matches the “meeting” uttered by the speaker B, the estimating unit 23 first estimates the same schedule. Then, the estimation unit 23 uses a plurality of different keywords “belonging to the same category in the extracted keywords (“ April ”,“ 1st ”,“ 2nd ”,“ meeting ”,“ 3 o'clock ”,“ meeting room ”). Since there are “1 day” and “2 days”, it is estimated that there is a schedule change.

  Subsequently, the selection unit 24 selects a keyword spoken by a plurality of speakers A and B in preference to a keyword spoken by a speaker alone, from “1 day” or “2 days”. That is, since “Speaker A” uttered a total of two times on “1 day” and “Speaker A” and “B” uttered twice each on “2 days”, the selection unit 24 “2 days” is selected as a keyword used for the sharing schedule. As described above, when the intentions of a plurality of speakers agree with each other rather than the speaker alone, the keyword selection is weighted. At the same time, the selection unit 24 selects one keyword (“April”, “Meeting”, “3 o'clock”, “Meeting room”) belonging to the same category. Then, the schedule creation unit 25 creates the shared schedule shown in FIG. 8 based on these keywords (“April”, “2nd day”, “meeting”, “3 o'clock”, “conference room”).

  In the case of the above example 2, when the speaker A speaks “1 day” once and the speaker B speaks “2 days” once, the selection unit 24 selects the keyword in time series later. “2 days” may be selected in preference to the keyword “1 day”. This is because a keyword that is uttered later in time series is highly likely to be a keyword updated in the conversation.

  Returning to FIG. 7, when it is determined that there are not a plurality of identified speakers (step S <b> 14; No), that is, when there is only one speaker, the utterance voice of the speaker alone in the manner of the estimation unit 23 and the selection unit 24 described above. A keyword is selected based on whether or not the schedule has been changed (step S17). In the case of a speaker alone, if there is a schedule change within the same schedule, the selection unit 24 selects, for example, a keyword having the largest number of utterances by a speaker among a plurality of keywords, or if the number of utterances is the same. A keyword is selected by selecting a later keyword in time series.

  Next, the schedule creation unit 25 creates a single schedule based on the selected keyword (step S18). After creating a shared schedule or a single schedule (after step S16 or S18), the schedule creation process ends.

  As described above, according to the schedule creation device 200 according to the present embodiment, by providing the estimation unit 23 and the selection unit 24, not only a shared schedule is created based on the extracted keyword, but also a conversation If there is a schedule change, a shared schedule can be created using keywords selected from two or more keywords in the same category. For this reason, it is possible to create an accurate sharing schedule that takes into account the schedule change in the conversation.

  The selection unit 24 selects a keyword uttered by a plurality of speakers in preference to a keyword uttered by a speaker alone from two or more keywords of the same category. For this reason, it is possible to create a more accurate sharing schedule in consideration of matching of intentions of a plurality of speakers. Note that the selection method of the selection unit 24 is not limited to this, and, for example, a later keyword in time series may be selected with priority over the preceding keyword. According to this, it is possible to create an accurate sharing schedule reflecting keywords updated in the conversation, for example, correction of mistakes.

  Moreover, the estimation part 23 estimates the presence or absence of the schedule change in conversation speech, excluding the schedule different from the schedule concerning the schedule change. According to this, since it is possible to avoid a situation in which a different schedule in the conversation is mistaken as a schedule change, an accurate sharing schedule can be created. Note that the schedule creation unit 25 may create two shared schedules at the same time using the excluded different schedule as a shared schedule.

  Further, in this embodiment, the criterion for determining whether or not the schedules are the same (a criterion to be excluded as a separate schedule) is whether or not the predetermined category “planned contents” matches. You may determine whether it is the same schedule based on the number. For example, if the three categories “time”, “location”, and “planned content” match, it may be determined that the schedules are the same.

Moreover, the extraction part 22 of the schedule creation apparatus 200 extracts the keyword relevant to a schedule from a speaker's voice by text matching. When keyword extraction is performed by voice matching, that is, when each keyword in the keyword table 32 is a phoneme waveform consisting of a set of phonemes instead of text, the keywords uttered for each speaker are not necessarily the same phoneme waveform. .
For this reason, according to the keyword extraction technique based on text matching, the individuality of the sound for each speaker is lost, so the accuracy of keyword extraction can be improved. However, it does not hinder keyword extraction by comparing voices.

  Further, according to the schedule creation device 200 in this embodiment, the speaker identification unit 21 identifies a speaker by (1) voice recognition (2) image recognition (3) terminal ID. For this reason, it is sufficient that the speaker can be identified by any one of the speaker identification methods, and therefore the identification accuracy can be improved as compared with the speaker identification using only one method. In particular, it is suitable when the face of the speaker does not enter the angle of view of the camera 14 of the user terminal 100. Further, the present invention is not limited to the above (1) to (3), and the speaker may be identified by comparing the speaker's fingerprint data or iris pattern.

  In the schedule creation process of FIG. 7 described above, if there are a plurality of identified speakers (step S14; Yes), a sharing schedule is created for all of the plurality of speakers. Absent. For example, after step S <b> 14, a step of determining whether or not a predetermined condition is satisfied among a plurality of speakers is added, and then the schedule creation unit 25 creates a shared schedule shared by the speakers satisfying the predetermined condition. You may create it.

  For example, volume or tone can be used as the predetermined condition. When the volume is used, when the speaker identification unit 21 identifies a plurality of speakers, the control unit 20 shares a sharing schedule among the plurality of speakers based on the volume of the conversation voices of the plurality of speakers. What is necessary is just to identify a speaker. And the schedule preparation part 25 should just create the sharing schedule shared with the speaker which the control part 20 specified. In this case, the control unit 20 functions as a specifying unit.

Specifically, the control unit 20 is a talk that shares a sharing schedule with a speaker whose conversation voice volume is equal to or lower than a predetermined threshold among a plurality of speakers identified by the speaker identification unit 21. What is necessary is just to identify as a person.
According to this aspect, a registered user (for example, speakers A and B in FIG. 1) who participates in a conference or briefing, and a passing person who is a registered user but is not a participant in the conference or briefing (for example, a story) A registered user who is not a registered user name C) can be determined based on the sound volume, and a sharing schedule shared only by the former can be created. This is because speakers who make schedule meetings in an environment where people other than participants in conferences and briefings pass are highly likely to speak quietly so that the content of the conversation can be heard only from one another.

In addition, when the volume of at least one of the conversation voices among the plurality of speakers identified by the speaker identification unit 21 is equal to or greater than a predetermined threshold, the control unit 20 includes a plurality of stories identified by the speaker identification unit 21. It is good also as specifying all the speakers as speakers who share a sharing schedule.
According to this aspect, it is possible to share and set a schedule to be made known to all participants in a meeting or briefing as a schedule shared by all participants. This is because if the speaker is speaking about the schedule louder than a certain level, there is a high possibility that all the participants participating in the conference or briefing want to share the schedule.

  Although the description of the embodiment has been completed above, the embodiment is an example, and it is needless to say that the configuration of the schedule device 200, the contents of the schedule creation process, and the like are not limited to those described in the embodiment.

(Second Embodiment)
In the first embodiment described above, the schedule creation device 200 registers a schedule without determining whether or not the schedule created by the schedule creation unit 25 has already been registered. Therefore, the schedule creation apparatus 200 ′ according to the second embodiment is different from the schedule creation apparatus 200 in that a new function for avoiding duplicate registration is added. Below, it demonstrates centering on this different point.

  As illustrated in FIG. 9, the schedule creation device 200 ′ includes a determination unit 26 and an update unit 27 as new functions.

  The determination unit 26 determines whether or not the sharing schedule created by the schedule creation unit 25 has been shared by all the speakers. The determination unit 26 determines whether or not the single schedule created by the schedule creation unit 25 has been registered.

  When the determination unit 26 determines that the sharing schedule has not been shared with all the speakers, the update unit 27 updates the schedule of speakers that have not been shared. Moreover, the update part 27 registers the single schedule which the schedule preparation part 25 created, when the determination part 26 determines with the single schedule not having been registered.

  Hereinafter, the schedule creation processing of the schedule creation device 200 'will be described with reference to FIG. Since this schedule creation process is the same as the schedule creation process according to the first embodiment except that steps S19 to S22 relating to the new function are added, the description will focus on different steps.

  After the schedule creation unit 25 creates a sharing schedule in step S16, the determination unit 26 determines whether or not the sharing schedule has been shared (step S19). Specifically, the determination unit 26 determines whether or not the sharing schedule is already shared with a plurality of speakers.

Here, when it is determined that the sharing schedule is not shared by all the speakers (step S19; No), the updating unit 27 updates the schedule of speakers who have not shared the sharing schedule (step S20). ), The process is terminated.
For example, when the identified speakers are A and B, and the existing schedule is registered only in A, the single schedule before update shown in FIG. 11 is updated to the shared schedule shown in FIG.
On the other hand, if it is determined that the sharing schedule has been shared (step S19; Yes), for example, if speakers A and B have already shared the schedule, the process ends.

On the other hand, after the schedule creation unit 25 creates a single schedule in step S18, the determination unit 26 determines whether the single schedule has been registered (step S21).
Here, when it is determined that the single schedule is not already registered (step S21; No), the update unit 27 newly registers the single schedule (step S22) and ends the process. On the other hand, when it is determined that the single schedule has already been registered (step S21; Yes), the process ends.

  As described above, according to the schedule creation device 200 ′ according to the second embodiment, when the determination unit 26 and the update unit 27 are provided, the shared schedule is not shared and the single schedule is not registered. Updates and new registrations can be made only with Therefore, according to the schedule creation device 200 ′, update / new registration can be performed only when necessary while avoiding duplicate registration.

(Third embodiment)
In the first and second embodiments described above, the schedule creation device 200 is described on the assumption that the speaker is identified using the voice, image, and terminal ID of the speaker received from the user terminal 100 which is another device. However, it is not limited to this. For example, as shown in FIG. 13, the schedule creation device 300, which is a notebook PC, may have a stand-alone configuration in which sound is collected and an image is captured.

  In this case, the schedule creation device 300 includes a microphone 50 for collecting sound and a camera 60 for capturing an image, as shown in FIG. Then, the schedule creation device 300 creates the schedule by executing the schedule creation processing of FIG. 7 described above based on the sound collected by the own device and the captured image. The schedule creation process may be triggered when the speaker A or B activates the schedule creation program and inputs a start instruction. Note that the created shared schedule can be referred to by anyone other than the owner of the schedule creation device 300 by uploading it to a server or the like.

  According to the schedule creation device 300 according to the third embodiment described above, since it is a stand-alone type, a schedule is created by quickly performing speaker identification and keyword extraction without causing a delay associated with transmission of voice and images. can do. For this reason, since a schedule can be created immediately regardless of the state of the communication network, usability can be improved.

(Fourth embodiment)
In the first embodiment described above, the schedule creating apparatus 200 has been described on the assumption that a shared schedule shared by all of a plurality of speakers is created. In this case, if a speaker who does not speak a keyword related to the schedule is also identified from the conversation voice, a sharing schedule is created. For example, a sharing schedule that is shared by ABC even when C other than speakers A and B does not speak a keyword is created.

  Therefore, the schedule creation unit 25 of the schedule creation device 400 according to the fourth embodiment shown in FIG. 15 creates a sharing schedule that is shared only by the speaker who spoke the keyword extracted by the extraction unit 22. In this case, for example, when the speaker identification unit 21 identifies a speaker using the speaker identification template, the speaker identification unit 21 identifies the speaker from the voiceprint of the utterance voice portion of the keyword extracted by the extraction unit 22. And the schedule preparation part 25 should just create the share schedule (The share schedule shared by AB except C in the above-mentioned example) shared only by the speaker who spoke the keyword.

  The schedule correction unit 28 corrects the shared schedule created by the schedule creation unit 25 and stored in the schedule storage unit 34. The schedule creation device 200 according to the first embodiment described above creates a corrected shared schedule reflecting the correction from the conversation voice received in real time by the estimation unit 23, the selection unit 24, and the like. Modify the shared schedule once stored. Although the correction method of the schedule correction unit 28 is arbitrary, for example, when a plurality of keywords are extracted in the same category by the same function as the estimation unit 23, the selection unit 24, etc. based on the recorded conversation voice Correction may be performed by selecting one keyword or the like. Of course, not only the recorded voice but also the once-stored shared schedule may be corrected in real time.

According to the schedule creation device 400 according to the fourth embodiment, it is possible to create a sharing schedule between speakers excluding speakers who do not need to share a schedule while improving the accuracy of the sharing schedule by a correction function. .
Multiple schedules may be shared. For example, when speaker ABCDE is speaking, the schedule of “April 2 Planning Meeting 3:00 Meeting Room” is shared by Speaker ABC, and “April 3 Sales Meeting” The schedule for the “3 o'clock meeting room” is shared by the speaker ACD.

In the first and second embodiments described above, the schedule creation processing in FIG. 7 starts the creation of the shared schedule if there are a plurality of identified speakers (step S14; Yes), but is not limited thereto. Absent. For example, after the step S14, a process for determining whether or not there is a speaker's consent may be added, and creation of a shared schedule may be started with the presence of the consent as a trigger.
Specifically, keywords related to consent (for example, “I understand” or “OK”) are memorized in advance, and a sharing schedule is created triggered by any speaker speaking the keyword. May be. According to this, it is possible to avoid a situation where the sharing schedule is forcibly created when the speaker does not want to share the schedule.

In addition, in 1st and 2nd embodiment mentioned above, although the user terminal 100 demonstrated on the assumption that it was a smart phone, it is not restricted to this. Any portable type may be used. For example, a notebook PC, a tablet terminal, or the like can be used. The schedule creation devices 200 and 200 ′ are not limited to servers, and any device can be used as long as it can withstand the load of schedule creation processing. For example, a PC can be used.
Moreover, the aspect of the schedule format 33 common to each embodiment is an example, and it is needless to say that another aspect (for example, a calendar format) may be adopted.

The functions of the schedule creation devices 200, 200 ′, 300, and 400 according to the present invention can also be implemented by a computer such as a normal PC.
Specifically, in the above embodiment, the schedule creation processing program performed by the schedule creation devices 200, 200 ′, 300, and 400 has been described as being stored in advance in the ROM of the storage unit 30. However, schedule creation processing programs are stored and distributed on computer-readable recording media such as flexible disks, CD-ROMs (Compact Disc Read Only Memory), DVDs (Digital Versatile Discs), and MOs (Magneto-Optical Discs). Then, by installing the program in the computer, a computer capable of realizing the above-described functions may be configured.

  As mentioned above, although preferable embodiment of this invention was described, this invention is not limited to the specific embodiment which concerns, This invention includes the invention described in the claim, and its equivalent range It is. Hereinafter, the invention described in the scope of claims of the present application will be appended.

(Appendix 1)
An extraction means for extracting keywords belonging to a category related to a schedule from conversational voices of a plurality of speakers;
Based on whether the extraction means has extracted a plurality of different keywords belonging to the same category, the estimation means for estimating the presence or absence of a schedule change in the conversation voice;
When the estimating means estimates that there is a schedule change, out of a plurality of different keywords belonging to the same category, a selecting means for selecting a keyword used for a shared schedule shared by the plurality of speakers;
Schedule creation means for creating the shared schedule based on the keyword selected by the selection means;
A schedule creation device comprising:

(Appendix 2)
The estimating means estimates the presence or absence of a schedule change in the conversation voice by excluding a schedule different from the schedule related to the schedule change;
The schedule creation device according to Supplementary Note 1, wherein

(Appendix 3)
The estimation means estimates that there is a schedule change in the conversation voice when the extraction means extracts a plurality of different keywords belonging to the same category,
When the estimation unit estimates that there is a schedule change, the selection unit prioritizes a keyword spoken by the plurality of speakers over a keyword spoken by the speaker alone among a plurality of different keywords belonging to the same category. Select
The schedule creation device according to Supplementary Note 1 or 2, characterized in that:

(Appendix 4)
The estimation means estimates that there is a schedule change in the conversation voice when the extraction means extracts a plurality of different keywords belonging to the same category,
The selection means, when the estimation means estimates that there is a schedule change, out of a plurality of different keywords belonging to the same category, to select the subsequent keywords in time series with priority over the previous keywords,
The schedule creation device according to Supplementary Note 1 or 2, characterized in that:

(Appendix 5)
If the extraction means does not extract a plurality of different keywords belonging to the same category, the estimation means estimates that there is no schedule change in the conversation voice,
The selecting means selects one keyword belonging to the same category extracted by the extracting means when the estimating means estimates that there is no schedule change;
The schedule creation device according to any one of supplementary notes 1 to 4, characterized in that:

(Appendix 6)
Based on the volume or tone of the conversational voices of the plurality of speakers, comprising a specifying means for specifying a speaker sharing the sharing schedule among the plurality of speakers,
The schedule creating means creates a sharing schedule to be shared by the speakers identified by the identifying means based on the keywords selected by the selecting means;
The schedule creation device according to any one of appendices 1 to 5, characterized in that:

(Appendix 7)
Determining means for determining whether the sharing schedule created by the schedule creating means has been shared by all the plurality of speakers;
An update unit that updates a schedule of speakers that have not been shared when the determination unit determines that the plurality of speakers have not been shared;
The schedule creation device according to any one of supplementary notes 1 to 6, further comprising:

(Appendix 8)
The extraction means converts the conversation voice into text, and when the converted text matches the text of a keyword belonging to any of a plurality of categories learned in advance, the matched text is Extract as keywords,
The schedule creation device according to any one of supplementary notes 1 to 7, characterized in that:

(Appendix 9)
An identification means for identifying the plurality of speakers by voice recognition or image recognition;
The schedule creation device according to any one of appendices 1 to 8, characterized in that:

(Appendix 10)
Speaker identification means for identifying the speaker;
When the speaker identification means identifies a plurality of speakers, an extraction means for extracting keywords belonging to a category related to the schedule from the conversation voices of the plurality of speakers;
Schedule creation means for creating a sharing schedule that is shared only by the speaker who spoke the keyword extracted by the extraction means;
Schedule correction means for correcting the sharing schedule created by the schedule creation means based on conversational voices of the plurality of speakers;
A schedule creation device comprising:

(Appendix 11)
The schedule creation means creates a shared schedule that is shared only by speakers who have spoken a common keyword among the keywords extracted by the extraction means.
The schedule creation device according to Supplementary Note 10, wherein

(Appendix 12)
An extraction step of extracting keywords belonging to a category related to the schedule from conversational voices of a plurality of speakers;
Based on whether or not a plurality of different keywords belonging to the same category are extracted in the extraction step, an estimation step for estimating the presence or absence of a schedule change in the conversation voice;
When it is estimated that there is a schedule change in the estimation step, a selection step of selecting a keyword to be used for a sharing schedule shared by the plurality of speakers among a plurality of different keywords belonging to the same category;
A schedule creation step of creating the shared schedule based on the keyword selected in the selection step;
A schedule creation method characterized by comprising:

(Appendix 13)
Computer
An extraction means for extracting keywords belonging to a category related to a schedule from conversational voices of a plurality of speakers;
Estimating means for estimating the presence or absence of a schedule change in the conversation voice based on whether or not the extracting means has extracted a plurality of different keywords belonging to the same category;
A selection means for selecting a keyword to be used for a shared schedule shared by the plurality of speakers from a plurality of different keywords belonging to the same category when the estimation means estimates that there is a schedule change;
Schedule creation means for creating the shared schedule based on the keyword selected by the selection means;
Program to function as.

  DESCRIPTION OF SYMBOLS 100 ... User terminal, 200, 200 ', 300, 400 ... Schedule creation apparatus, 11 ... Control part, 12 ... Input part, 13, 50 ... Microphone, 14, 60 ... Camera, 15 ... Memory | storage part, 16 ... Communication part, DESCRIPTION OF SYMBOLS 17 ... Display part, 20 ... Control part, 21 ... Speaker identification part, 22 ... Extraction part, 23 ... Estimation part, 24 ... Selection part, 25 ... Schedule preparation part, 26 ... Determination part, 27 ... Update part, 28 ... Schedule correction unit, 30 ... storage unit, 31 ... speaker identification template, 32 ... keyword table, 33 ... schedule format, 34 ... schedule storage unit, 40 ... communication unit

Claims (13)

An extraction means for extracting keywords belonging to a category related to a schedule from conversational voices of a plurality of speakers;
Based on whether the extraction means has extracted a plurality of different keywords belonging to the same category, the estimation means for estimating the presence or absence of a schedule change in the conversation voice;
When the estimating means estimates that there is a schedule change, out of a plurality of different keywords belonging to the same category, a selecting means for selecting a keyword used for a shared schedule shared by the plurality of speakers;
Schedule creation means for creating the shared schedule based on the keyword selected by the selection means;
A schedule creation device comprising: The estimating means estimates the presence or absence of a schedule change in the conversation voice by excluding a schedule different from the schedule related to the schedule change;
The schedule creation device according to claim 1. The estimation means estimates that there is a schedule change in the conversation voice when the extraction means extracts a plurality of different keywords belonging to the same category,
When the estimation unit estimates that there is a schedule change, the selection unit prioritizes a keyword spoken by the plurality of speakers over a keyword spoken by the speaker alone among a plurality of different keywords belonging to the same category. Select
The schedule creation device according to claim 1 or 2, wherein The estimation means estimates that there is a schedule change in the conversation voice when the extraction means extracts a plurality of different keywords belonging to the same category,
The selection means, when the estimation means estimates that there is a schedule change, out of a plurality of different keywords belonging to the same category, to select the subsequent keywords in time series with priority over the previous keywords,
The schedule creation device according to claim 1 or 2, wherein If the extraction means does not extract a plurality of different keywords belonging to the same category, the estimation means estimates that there is no schedule change in the conversation voice,
The selecting means selects one keyword belonging to the same category extracted by the extracting means when the estimating means estimates that there is no schedule change;
The schedule creation device according to claim 1, wherein the schedule creation device is a device for creating a schedule. Based on the volume or tone of the conversational voices of the plurality of speakers, comprising a specifying means for specifying a speaker sharing the sharing schedule among the plurality of speakers,
The schedule creating means creates a sharing schedule to be shared by the speakers identified by the identifying means based on the keywords selected by the selecting means;
The schedule creation device according to claim 1, wherein the schedule creation device is a device for creating a schedule. Determining means for determining whether the sharing schedule created by the schedule creating means has been shared by all the plurality of speakers;
An update unit that updates a schedule of speakers that have not been shared when the determination unit determines that the plurality of speakers have not been shared;
The schedule creation device according to any one of claims 1 to 6, further comprising: The extraction means converts the conversation voice into text, and when the converted text matches the text of a keyword belonging to any of a plurality of categories learned in advance, the matched text is Extract as keywords,
The schedule creation device according to claim 1, wherein the schedule creation device is a device for creating a schedule. An identification means for identifying the plurality of speakers by voice recognition or image recognition;
The schedule creation device according to claim 1, wherein the schedule creation device is a device for creating a schedule. Speaker identification means for identifying the speaker;
When the speaker identification means identifies a plurality of speakers, an extraction means for extracting keywords belonging to a category related to the schedule from the conversation voices of the plurality of speakers;
Schedule creation means for creating a sharing schedule that is shared only by the speaker who spoke the keyword extracted by the extraction means;
Schedule correction means for correcting the sharing schedule created by the schedule creation means based on conversational voices of the plurality of speakers;
A schedule creation device comprising: The schedule creation means creates a shared schedule that is shared only by speakers who have spoken a common keyword among the keywords extracted by the extraction means.
The schedule creation device according to claim 10. An extraction step of extracting keywords belonging to a category related to the schedule from conversational voices of a plurality of speakers;
Based on whether or not a plurality of different keywords belonging to the same category are extracted in the extraction step, an estimation step for estimating the presence or absence of a schedule change in the conversation voice;
When it is estimated that there is a schedule change in the estimation step, a selection step of selecting a keyword to be used for a sharing schedule shared by the plurality of speakers among a plurality of different keywords belonging to the same category;
A schedule creation step of creating the shared schedule based on the keyword selected in the selection step;
A schedule creation method characterized by comprising: Computer
An extraction means for extracting keywords belonging to a category related to a schedule from conversational voices of a plurality of speakers;
Estimating means for estimating the presence or absence of a schedule change in the conversation voice based on whether or not the extracting means has extracted a plurality of different keywords belonging to the same category;
A selection means for selecting a keyword to be used for a shared schedule shared by the plurality of speakers from a plurality of different keywords belonging to the same category when the estimation means estimates that there is a schedule change;
Schedule creation means for creating the shared schedule based on the keyword selected by the selection means;
Program to function as.

Images