JP7243145B2 - Information processing device, information processing system and information processing method - Google Patents

Description

The present invention relates to an information processing device, an information processing system, and an information processing method.

When manually creating minutes during a meeting, omissions, omissions, or individual subjectivity may be entered. On the other hand, recording conference voices and transcribing them into characters at a later date to create minutes requires a great deal of labor and time. Therefore, in order to reduce the labor required to create minutes, there is known a meeting support system that performs voice recognition during a meeting in real time and converts it into text to support the creation of minutes without omissions or omissions during the meeting. (For example, Patent Document 1).

However, conventional conference support systems that use speech recognition require high-performance microphones or microphones with excellent sound-collecting properties in order to improve the accuracy of speech recognition. rice field.

In addition, when conference participants individually recognize their voices with microphones, the voice recognition is performed with each microphone, so that the conference support system may acquire the speaker's utterances redundantly. In addition, it is difficult to obtain the minutes correctly when the same speech is repeated or when a plurality of conference participants speak at the same time. On the other hand, the sound-insulating microphone used to capture the voice of only the speaker is an expensive and complicated device, making the operation of the conference support system difficult.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and it is an object of the present invention to process the speech as one person's utterance even when the speech is simultaneously acquired by a plurality of microphones from the same sound source.

Therefore, in order to solve the above problem, an information processing apparatus is connected to a plurality of information processing terminals, and acquisition of the text and the voice obtained by recognizing the voice acquired by the information processing terminal is started. an acquisition unit that acquires a message including a time, text included in a first message acquired by the acquisition unit, and the time before the time included in the first message acquired by the acquisition unit; a first calculator that calculates a degree of similarity with a text included in a second message; and whether or not to record the first message based on the degree of similarity calculated by the first calculator and a second calculation unit for calculating the similarity of the speech waveform of the speech, and when it is determined that the similarity is similar by the similarity calculated by the first calculation unit , the second calculation unit calculates a similarity between a speech waveform included in the first message and a speech waveform included in the second message, and the determination unit performs the second calculation unit It is determined whether to record the first message based on the similarity calculated by .

Even when voices are simultaneously acquired from the same sound source by a plurality of microphones, they can be processed as utterances by one person.

It is a figure showing an example of composition of information processing system 1 in an embodiment of the invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a figure for demonstrating the information processing system 1 in embodiment of this invention. It is a figure which shows the hardware structural example of the information processing apparatus 10 in embodiment of this invention. 1 is a diagram illustrating an example of functional configuration of an information processing apparatus 10 according to a first embodiment; FIG. It is a figure for demonstrating the minimum edit distance algorithm in 1st Embodiment. FIG. 7 is a diagram showing an example of a method of calculating similarity between speech waveforms according to the first embodiment; 4 is a flowchart for explaining an information processing method according to the first embodiment; It is a figure which shows the functional structural example of the information processing apparatus 10 in 2nd Embodiment. FIG. 12 is a sequence diagram for explaining example (1) of information processing in the second embodiment; FIG. 12 is a sequence diagram for explaining example (2) of information processing in the second embodiment; It is a figure which shows the example (1) of the conventional information processing. FIG. 10 is a diagram illustrating an example (1) of information processing in the second embodiment; It is a figure which shows the example (2) of the conventional information processing. FIG. 10 is a diagram illustrating an example (2) of information processing in the second embodiment;

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below based on the drawings.

FIG. 1 is a diagram showing a configuration example of an information processing system 1 according to an embodiment of the present invention. As shown in FIG. 1 , an information processing system 1 according to the embodiment of the present invention has an information processing device 10 and a client 20 . The information processing system 1 is, for example, a conference support system that can input conference audio and output minutes in real time.

The information processing device 10 is a computer such as a server, and is a device that implements functions by a program described later. As another example, the information processing device 10 may be a conference system dedicated device, an electronic whiteboard, a projector, or the like. As shown in FIG. 1, the information processing apparatus 10 receives at least a speech waveform from each client 20 and a text message that is the result of speech recognition of the speech waveform by the client 20, and filters the text message. Send the message to the client 20 . Note that the information processing apparatus 10 may receive only a voice waveform from each client and perform voice recognition of the voice waveform. Furthermore, the information processing apparatus 10 may have a function of displaying statistics of the number of participants and the number of statements as information incidental to the creation of minutes.

The client 20 is a computer such as a PC (Personal Computer) including at least a microphone and a display device. The client 20 includes multiple information processing terminals. As another example, the client 20 may be a tablet terminal, smart phone, PDA (Personal Digital Assistant), mobile phone, wearable PC, game device, car navigation terminal, or the like.

FIG. 2 is a diagram for explaining the information processing system 1 according to the embodiment of the present invention. The information processing system 1 according to the embodiment of the present invention can provide a real-time minutes support service. The real-time minutes support service is a service that converts remarks made in a conference into text by speech recognition, generates minutes, and provides real-time feedback to conference participants.

As shown in FIG. 2, in the information processing system 1, it is possible to record and edit comments in a chat format. It is also possible to display the statistics of the number of participants and the number of remarks. In addition, voice recognition input can be used, and an external engine can be used for voice recognition. In addition, the check function can support issuing minutes. The check function is, for example, a function of marking contents judged to be important during the meeting and supporting review after the meeting. Synchronization to a user-configured business chat service is also envisioned. Also, at the end of the meeting, the summary based on the tagging information and the acquired raw data can be packed and exported to e-mail or the like. Tagging information may include, for example, keywords. Summarization may be based on utterances containing keywords.

It should be noted that the screen shown in FIG. 2 is a screen on the client 20 installed for each conference participant. can be viewed and commented on the PC.

FIG. 3 is a diagram showing a hardware configuration example of the information processing device 10 according to the embodiment of the present invention. The information processing apparatus 10 shown in FIG. 3 has a processor 11, a storage device 12, an auxiliary storage device 13, an input/output interface 14, a communication interface 15, etc., which are connected to each other.

A program that implements processing in the information processing device 10 is stored in the auxiliary storage device 13 . The auxiliary storage device 13 holds installed programs and necessary data.

The storage device 12 reads the program from the auxiliary storage device 13 and stores it in accordance with the activation instruction from the processor 11 . The processor 11 realizes functions related to the information processing apparatus 10 by executing programs stored in the storage device 12 .

The input/output interface 14 is an interface for connecting various input/output devices such as a microphone, a USB (Universal Serial Bus) device, a hardware key, a status notification LED, and a liquid crystal display.

The communication interface 15 is a wired or wireless interface for communicating with the client 20, mail server, smart phone, PC, or the like.

Note that the client 20 may also have the same hardware configuration as in FIG. Client 20 may also have a microphone connected to input/output interface 14 .

A first embodiment of the present invention will be described below. FIG. 4 is a diagram showing a functional configuration example of the information processing apparatus 10 according to the first embodiment. As shown in FIG. 4, the information processing apparatus 10 has an information receiving unit 101, a message storage unit 102, a message similarity calculation unit 103, a voice storage unit 104, a voice similarity calculation unit 105, and a similar message determination unit 106. . Each functional unit of the information processing apparatus 10 is implemented by the processor 11 that executes a program developed on the storage device 12 from the auxiliary storage device 13 shown in FIG.

The information receiving unit 101 is realized by the communication interface 15 shown in FIG. 3, and receives at least a speech waveform and a text message that is the result of speech recognition of the speech waveform by the client 20 from each client 20 . Information indicating the microphone ID from which the voice was acquired and the time at which the voice acquisition started is added to the voice waveform and the text message. It should be noted that the time of each client 20 may be synchronized with a predetermined accuracy, and the time at which the voice was acquired may be measured at the synchronized time. The microphone ID may be information identifying the client 20 or information identifying the microphone itself.

The message storage unit 102 receives and stores the text message from the information reception unit 101 . Text messages received from the client 20 during a predetermined period are stored in the message storage unit 102 . The message storage unit 102 may store the text message, the associated microphone ID, and the time when the acquisition of the voice is started in association with each other.

The message similarity calculation unit 103 determines whether the received text message is similar to the text messages stored in the message storage unit 102 using, for example, a minimum edit distance algorithm.

FIG. 5 is a diagram for explaining the minimum edit distance algorithm in the first embodiment. A minimum edit distance algorithm is an algorithm that compares the degree of similarity between characters. As shown in FIG. 5, the distance is defined as the minimum number of times that a character string can be matched with another character string by performing three processes of "insertion", "delete" or "replacement". Define. It is determined that the smaller the distance between character strings, the higher the degree of similarity.

"Insert" shown in FIG. 5 is an example in which two characters "bun" are inserted into "shinbunshi" to form "shinbunshi". The distance between "shinbunshi" and "shinbunshi" is two, because two characters, that is, are "inserted" twice.

"Deletion" shown in FIG. 5 is an example in which "shi" at the end of "shinshi" is deleted to become "shin". Since one character is "deleted" once, the distance between "shin" and "shin" is one.

"Replacement" shown in FIG. 5 is an example in which "n" of "shinshi" is replaced with "ka" to form "but". The distance between "shinshi" and "but" is 1 because there is one character, that is, one "replacement".

Return to FIG. The message similarity calculation unit 103 determines whether or not the newly received text message is similar to the stored text message, and the result is sent to the similar message determination unit 106 . Note that the minimum edit distance algorithm used to determine similarity is an example, and other algorithms may be used to determine similarity of text messages.

The voice storage unit 104 receives and stores voice waveforms from the information receiving unit 101 . A voice waveform received from the client 20 during a predetermined period is stored in the voice storage unit 104 . The voice storage unit 104 may store the voice waveform, the associated microphone ID, and the time at which voice acquisition is started in association with each other.

The speech similarity calculation unit 105 determines whether the received speech waveform is similar to the speech waveform stored in the speech storage unit 104, for example, by frequency analysis and using a cross-correlation function.

FIG. 6 is a diagram showing an example of a method of calculating the similarity of speech waveforms according to the first embodiment. As shown in FIG. 6, this is a method of frequency-converting the voice waveform and calculating the degree of similarity using the cross-correlation function. However, the method below is an example of general frequency analysis, and the calculation of the similarity of speech waveforms is not limited to the algorithm below.

The right figure in FIG. 6 is obtained by Fourier transforming the speech waveform in the left figure in FIG. For example, if the speech waveforms for which the similarity is to be calculated are c(t) and d(t), and the Fourier transforms of them are C(k) and D(k), then the following numbers 1, the similarity of speech waveforms is calculated. The greater the X in Equation 1, the higher the degree of similarity.

Return to FIG. The speech similarity calculation unit 105 determines whether or not the newly received speech waveform is similar to the accumulated speech waveform, and the result is sent to the similar message determination unit 106 . Note that the method of using the cross-correlation function described in FIG. 6 for similarity determination is an example, and the similarity of speech waveforms may be determined using other algorithms.

The similar message determination unit 106 transmits the filtered message to the client 20 based on the determination results received from the message similarity calculation unit 103 and the speech similarity calculation unit 105 . Details will be described later.

FIG. 7 is a flow chart for explaining the information processing method according to the first embodiment. Referring to FIG. 7, processing when a new message is acquired from the client 20 in the information processing apparatus 10 will be described.

In step S<b>11 , the information receiving section 101 acquires a new message from the client 20 . The new message includes information indicating a voice waveform, a text message that is the result of voice recognition of the voice waveform by the client 20, a microphone ID from which the voice was acquired, and a voice acquisition start time. Additionally, the new message may include an identifier to identify the message. Also, the new message is the latest message. That is, the new message has a later voice acquisition start time than any of the messages stored in message storage unit 102 or voice storage unit 104 .

In the subsequent step S12, the message similarity calculation unit 103 refers to the text messages stored in the message storage unit 102, and determines whether there is a text message that is most recently similar to the text message included in the new message ( S13). The term “most recent” is set to a period longer than a period during which messages may be acquired from the same sound source by a plurality of microphones. For example, if the maximum distance between microphones arranged in a conference room is 20 m, a period longer than the assumed maximum delay of 20 _m /340 _m/s = 0.0588 s may be set to 60 ms. 340 _m/s is an example of the speed of sound. The message similarity calculator 103 determines the similarity of a message having a voice acquisition start time within 60 ms from the voice acquisition start time included in the new message to the text message.

If the message was recently retrieved, the similarity between the text message of the new message and the stored text message is calculated using, for example, the minimum edit distance algorithm described in FIG. A plurality of text messages may be used for similarity determination. The message similarity calculation unit 103 determines whether or not the text messages are similar based on whether or not the minimum edit distance between text messages is less than a predetermined threshold. If they are similar (YES in S13), the process proceeds to step S14, and if they are not similar (NO in S13), the process proceeds to step S18.

In step S14, the message similarity calculation unit 103 refers to the voice log stored in the voice storage unit 104 of the message determined to be similar, and determines whether the voice waveform contained in the new message has a high degree of similarity. judge. For example, the similarity between the new speech waveform and the stored speech waveform is calculated using the method for calculating the speech waveform similarity described in FIG. A plurality of speech waveforms may be used for similarity determination. The message similarity calculator 103 determines whether or not the similarity between voice waveforms is similar based on whether or not the similarity between the voice waveforms is less than a predetermined threshold. If similar (YES in S15), proceed to step S16; otherwise (NO in S15), proceed to step S18.

In step S16, the volume of the past message determined to have high similarity in voice waveform in step S15 is compared with the volume of the new message. If the volume of the new message is high (YES in S16), the process proceeds to step S17. If the past message volume is high (NO in S16), the process proceeds to step S19.

In step S17, a removal flag is set in the past message. The past messages with the removal flag set stored in the message storage unit 102 and voice storage unit 104 will not be used for similarity determination thereafter.

In step S18, all clients 20 are notified of new messages and past messages with removal flags set. In the client 20, when a message with a "removal flag" set is notified, the message may not be displayed on the screen. In step S18, the new message may be notified to the client 20 of the identifier of the message, the microphone ID and some or all of the text message. In step S18, for the past message with the removal flag set, only the message identifier and removal flag may be notified to the client 20, or the microphone ID and part or all of the text message may be notified to the client 20. good. In addition, the client 20 may not be notified of the past message with the removal flag set. The notified new message is recorded in the minutes.

In step S19, a removal flag is set for the new message. New messages with the removal flag set stored in the message storage unit 102 and voice storage unit 104 may not be stored.

In step S20, all clients 20 are notified of new messages and past messages with removal flags set. In the client 20, when a message with a "removal flag" set is notified, the message may not be displayed on the screen. In step S20, the new message may be notified to the client 20 of the identifier of the message, the microphone ID and some or all of the text message. For a new message with a removal flag set, only the message identifier and removal flag may be notified to the client 20, or the microphone ID and part or all of the text message may be notified. Also, the client 20 may not be notified of the new message with the removal flag set. New messages that do not have the posted removal flag set are logged in the minutes.

In the flowchart described in FIG. 7, if there is a message that is most recently similar to the new message in step S13 (YES in S13), the process may proceed to step S19 to set the removal flag for the new message.

As described above, according to the first embodiment, the information processing apparatus 10 compares the text message similarity between the latest new message obtained from each client 20 and the previously obtained message. If there are messages with similar text messages, the information processing apparatus 10 further compares the similarities of the voice waveforms. If there are messages with similar voice waveforms, the information processing apparatus 10 further compares the volume of the voice waveforms and records the message with the loudest volume in the minutes.

That is, even when voices are simultaneously acquired from the same sound source by a plurality of microphones, they can be processed as utterances by one person. Therefore, the readability of the minutes generated in the conference support system is improved, and the information processing apparatus 10 can remove messages that become noise due to voice input to multiple microphones provided in the client 20. , hardware such as expensive microphones becomes unnecessary.

Next, a second embodiment of the invention will be described. 2nd Embodiment demonstrates a different point from 1st Embodiment. Therefore, points that are not particularly mentioned may be the same as those in the first embodiment.

FIG. 8 is a diagram showing a functional configuration example of the information processing apparatus 10 according to the second embodiment. As shown in FIG. 8, the information processing apparatus 10 has an information receiving section 111, a duplicate message candidate storage section 112, an all message storage section 113, a message similarity determination section 114, and a message filtering section 115. FIG. Each functional unit of the information processing apparatus 10 is implemented by the processor 11 that executes a program developed on the storage device 12 from the auxiliary storage device 13 shown in FIG.

Information receiving section 111 has a function of extracting messages that may overlap in the time domain, in addition to the function of information receiving section 101 shown in FIG. The information receiving unit 110 transmits the extracted potentially duplicate messages to the duplicate message candidate storage unit 112 . For example, the period from the time audio acquisition started to the time audio acquisition completed for one message is the same as the period from the time audio acquisition started to the time audio acquisition completed for another message. If duplicated, it is extracted as a possible duplicate message. Information receiving section 110 also transmits all messages to all message storage section 113 .

The duplicate message candidate storage unit 112 receives and stores text messages from the information reception unit 110 . Of the text messages received from the client 20 during a predetermined period, messages that may overlap in the time domain are stored in the duplicate message candidate storage unit 112 . In the example shown in FIG. 8, message 2 and message 3 are messages that may overlap in the time domain. The duplicate message candidate storage unit 112 may associate and record the text message, the associated microphone ID, and the time when the acquisition of the voice was started and the time when the acquisition was completed.

The all message storage unit 113 receives and stores the text messages from the information reception unit 110 . Text messages received from the client 20 are stored in the all message storage unit 113 . The all-message storage unit 113 may associate and record the text message, the associated microphone ID, and the time when voice acquisition started and acquisition completion time. The all-message storage unit 113 transmits messages for which the filter is not set to ON to all clients.

The message similarity determination unit 114 has the functions of the message similarity calculation unit 103, the voice similarity calculation unit 105, and the similar message determination unit 106 shown in FIG. That is, the message similarity determination unit 114 can determine whether or not a certain message is similar to another message based on the text message and voice data. The message similarity determination unit 114 transmits the result of determining whether or not the messages acquired from the duplicate message candidate storage unit 112 or the all message storage unit 113 are similar to the message filtering unit 115 .

The message filtering unit 115, based on the result of determining whether the messages acquired from the message similarity determining unit 114 are similar and the result of determining whether the times at which the messages were acquired overlap, Has the ability to add filters. The message filtering unit 115 sends a notification to add the filter to the message to the all message storage unit 113, and the all message storage unit 113 turns on the filter for the target message. In the example shown in FIG. 8, the filter for message 3 is set on and message 3 is not sent to the client.

FIG. 9 is a sequence diagram for explaining example (1) of information processing in the second embodiment. In step S21, the client 20A notifies the information processing apparatus 10 that speech recognition has started. Subsequently, the information processing apparatus 10 records the time when the client 20A started speech recognition, and recognizes that speech recognition is in progress (S22). Subsequently, in step S23, the client 20A notifies the information processing apparatus 10 of completion of voice recognition and a message. The information processing apparatus 10 records the time when the client 20A finishes speech recognition, and recognizes that the speech recognition has finished. Subsequently, the information processing device 10 notifies all the clients 20 of the message (S24).

FIG. 10 is a sequence diagram for explaining example (2) of information processing in the second embodiment. In step S31, the client 20A notifies the information processing apparatus 10 that voice recognition has started. Subsequently, the information processing apparatus 10 records the time when the client 20A started speech recognition, and recognizes that speech recognition is in progress (S32). In step S33, the client 20B notifies the information processing apparatus 10 that speech recognition has started. Subsequently, the information processing apparatus 10 records the time when the client 20A started speech recognition, recognizes that speech recognition is in progress, and recognizes that the speech recognition times of the clients 20A and 20B overlap. presence is detected (S34).

In step S<b>35 , the client 20</b>B notifies the information processing apparatus 10 that the voice recognition has ended and the message B. The information processing apparatus 10 records the time when the client 20B finished speech recognition, recognizes that the speech recognition has finished, and confirms that the duplication of the speech recognition time between the client 20A and the client 20B is eliminated. can be detected. In step S36, the client 20A notifies the information processing apparatus 10 of the fact that the voice recognition has ended and the message A. FIG. The information processing apparatus 10 records the time when the client 20A finishes speech recognition, and recognizes that the speech recognition has finished. Here, the information processing apparatus 10 recognizes the period during which the message A was acquired and the period during which the message B was acquired. In step S37, the information processing apparatus 10 determines whether message A and message B overlap in the time domain based on the period during which message A and the period during which message B were obtained, which were obtained in steps S32 and S34. Determine whether the message is a possible message. If there is a possibility that message A and message B overlap, the information processing apparatus 10 performs processing similar to that of the flowchart shown in FIG. 7 to add a filter to the message with the removal flag set. On the other hand, the information processing apparatus 10 does not add a filter to message A and message B when there is no possibility that message A and message B overlap. In step S38, the information processing apparatus 10 notifies all the clients 20 of the unfiltered message and records it in the minutes.

In the information processing shown in FIG. 10, the number of messages to be processed is not limited to two, and three or more messages may be processed. If there are three or more messages for which there is a possibility that the acquired period overlaps, the same processing as in the flowchart shown in FIG. 7 is performed in step S37 to add a filter to the messages with the removal flag set. .

FIG. 11 is a diagram showing an example (1) of conventional information processing. As shown in FIG. 11, in a system that recognizes the voice of an individual with each microphone, it is possible to output an appropriate message based on the voice recognition result when the speech of the same person is acquired simultaneously with a plurality of microphones. Can not.

FIG. 12 is a diagram illustrating example (1) of information processing in the second embodiment. As shown in FIG. 12, the times when speech recognition overlaps are considered to determine potentially overlapping messages. By applying the information processing method shown in FIG. 7 to messages that may be duplicated, it is possible to prevent voice turbidity with multiple microphones.

FIG. 13 is a diagram showing an example (2) of conventional information processing. As shown in FIG. 13, when the similarity of speech recognition results is not taken into account and when different people make statements at the same time, one statement is adopted, so different statements can be output as messages. Can not.

FIG. 14 is a diagram illustrating example (2) of information processing in the second embodiment. By considering the degree of similarity of speech recognition results and the time when speech recognition overlaps, correct messages can be output to each person even if different people speak at the same time.

As described above, according to the second embodiment, the information processing apparatus 10 compares the text message similarity between the latest new message obtained from each client 20 and the previously obtained message. If there are messages with similar text messages, the information processing apparatus 10 further compares the similarities of the voice waveforms. If there are messages with similar voice waveforms, the information processing apparatus 10 further compares the volume of the voice waveforms and records the message with the loudest volume in the minutes. Furthermore, by filtering messages based on the time when speech recognition overlaps, even if different people speak at the same time, correct messages can be output to each person.

In addition, in the embodiment of the present invention, the information receiving section 101 is an example of an obtaining section. The message similarity calculator 103 is an example of a first calculator. Similar message determination unit 106 is an example of a determination unit. The speech similarity calculator 105 is an example of a second calculator. A microphone included in the client 20 is an example of a voice acquisition unit. The display device included in the client 20 is an example of a display unit.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to such specific embodiments, and various modifications can be made within the scope of the gist of the invention described in the claims.・Changes are possible.

10 information processing device 11 processor 12 storage device 13 auxiliary storage device 14 input/output interface 15 communication interface 20 information processing terminal (client)
101 information receiving unit 102 message storage unit 103 message similarity calculation unit 104 voice storage unit 105 voice similarity calculation unit 106 similar message determination unit 111 information reception unit 112 duplicate message candidate storage unit 113 all message storage unit 114 message similarity determination unit 115 message filtering unit

Japanese Patent Application Laid-Open No. 2006-301223

Claims (10)

An information processing device connected to a plurality of information processing terminals,
an acquisition unit configured to acquire a text obtained by recognizing a voice acquired by an information processing terminal and a message including a time when acquisition of the voice is started;
Text included in the first message acquired by the acquisition unit and text included in the second message acquired by the acquisition unit and having the time before the time included in the first message a first calculator that calculates the degree of similarity;
a determination unit that determines whether to record the first message based on the similarity calculated by the first calculation unit;
a second calculator that calculates the similarity of the speech waveform of the speech;
If the degree of similarity calculated by the first calculation unit determines that the similarity is similar, the second calculation unit calculates the voice waveform included in the first message and the voice waveform included in the second message. Calculate the similarity with the speech waveform that is
The information processing apparatus, wherein the determination unit determines whether to record the first message based on the degree of similarity calculated by the second calculation unit. 2. The information processing apparatus according to claim 1, wherein said time included in said first message and said time included in said second message are included in a predetermined period. 2. The information processing apparatus according to claim 1, wherein when the similarity calculated by the first calculation unit determines that the similarity is not similar, the determination unit determines to record the first message. 2. The information processing apparatus according to claim 1, wherein, when the degree of similarity calculated by the second calculation unit determines that the similarity is not similar, the determination unit determines to record the first message. If the degree of similarity calculated by the second calculation unit determines that the similarity is similar, the volume is determined based on the voice waveform included in the first message and the voice waveform included in the second message. and when the volume of the first message is louder, the decision unit decides to record the first message and delete the second message.
The information processing device described. If the degree of similarity calculated by the second calculation unit determines that the similarity is similar, the volume is determined based on the voice waveform included in the first message and the voice waveform included in the second message. 5 . The information processing apparatus according to claim 4 , wherein when the volume of the second message is louder, the decision unit decides to record the second message and delete the first message. The acquisition unit acquires a time when the acquisition of the voice of the first message is finished and a time when the acquisition of the voice of the second message is finished, and obtains a period during which the first message is obtained and obtain a time period during which the second message was obtained;
If the period during which the first message was acquired and the period during which the second message was acquired overlap, the determining unit determines, based on the degree of similarity calculated by the first calculating unit 2. The information processing apparatus according to claim 1, wherein the information processing apparatus determines whether or not to record the first message. The acquisition unit acquires a time when the acquisition of the voice of the first message is finished and a time when the acquisition of the voice of the second message is finished, and obtains a period during which the first message is obtained and obtain a time period during which the second message was obtained;
The determining unit records the first message and the second message when the period during which the first message is acquired and the period during which the second message is acquired do not overlap. 2. The information processing apparatus according to claim 1, wherein An information processing system including an information processing device connected to a plurality of information processing terminals,
The information processing device is
an acquisition unit configured to acquire a text obtained by recognizing a voice acquired by an information processing terminal and a message including a time when acquisition of the voice is started;
Text included in the first message acquired by the acquisition unit and text included in the second message acquired by the acquisition unit and having the time before the time included in the first message a first calculator that calculates the degree of similarity;
determining whether or not to record the first message based on the degree of similarity calculated by the first calculating unit; a determination unit that transmits to the information processing terminal;
a second calculator that calculates the similarity of the speech waveform of the speech;
If the degree of similarity calculated by the first calculation unit determines that the similarity is similar, the second calculation unit calculates the voice waveform included in the first message and the voice waveform included in the second message. Calculate the similarity with the speech waveform that is
The determination unit determines whether to record the first message based on the similarity calculated by the second calculation unit;
The information processing terminal
an audio acquisition unit that acquires audio;
and a display for displaying text included in the message. An information processing method executed by an information processing device connected to a plurality of information processing terminals,
an acquisition procedure for acquiring a text obtained by recognizing a voice acquired by an information processing terminal and a message including a time at which acquisition of the voice is started;
a text included in the first message obtained by the obtaining procedure; and a second text obtained by the obtaining procedure and having the time before the time included in the first message.
A first calculation procedure for calculating the similarity with the text included in the message of
a decision procedure for deciding whether to record the first message based on the similarity calculated by the calculation procedure;
and a second calculation procedure for calculating the similarity of the speech waveform of the speech,
If the degree of similarity calculated by the first calculation procedure determines that the similarity is similar, the second calculation procedure calculates the speech waveform included in the first message and the speech waveform included in the second message. Calculate the similarity with the speech waveform that is
The information processing method, wherein the determination procedure determines whether or not to record the first message based on the degree of similarity calculated by the second calculation procedure .

Images