KR20200009556A - Apparatus and Method for managing text changed from voice in call - Google Patents

Abstract

 Disclosed are an apparatus and method for converting and managing voice in a call into text. According to an aspect, an apparatus for text-converting a call voice of a service subscriber and managing the converted text includes: a receiver configured to receive voice data of a call in which the service subscriber becomes a caller or a receiver; A classification unit which classifies the received voice data into outgoing voice data and received voice data, respectively; A conversion unit for converting the classified voice data into data of the outgoing text and the received text, respectively; A call text unit for generating a call text by dividing the outgoing text of the converted sender and the received text of the receiver and listing them in chronological order; And a providing unit for inquiring the generated call text and providing the generated call text to the service subscriber's terminal at the request of the service subscriber.

Description

Apparatus and Method for managing text changed from voice in call}

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a technique of voice call, to an apparatus and method for converting a conversational voice of a caller and a receiver during a call into text and managing the converted text.

In the past, the only way to save a call was to record the call. The user installed the recording application on the mobile terminal, and stored the call contents as a voice file by pressing the recording button of the application during the mobile call.

Here, in order to confirm the recorded contents, the user must listen to the recorded voice file from the beginning to the end. If there are several recorded voice files, the user must listen to all of the plurality of voice files until a desired content is found.

Therefore, in the case of a plurality of voice files, it is difficult to identify which voice file the user wants and in which position of the voice file.

Korea Patent Registration 10-0935524 (2009.12.28)

The present invention has been made to solve the above problems, and an object of the present invention is to provide an apparatus and method for converting a call voice between a caller and a receiver into text and managing the converted text.

According to an aspect, an apparatus for text-converting a call voice of a service subscriber and managing the converted text includes: a receiver configured to receive voice data of a call in which the service subscriber becomes a caller or a receiver; A classification unit which classifies the received voice data into outgoing voice data and received voice data, respectively; A conversion unit for converting the classified voice data into data of an outgoing text and a received text, respectively; A call text unit for generating a call text by dividing the outgoing text of the converted sender and the received text of the receiver and arranging them in chronological order; And a provision unit for querying the generated call text and providing the generated call text to a terminal of the service subscriber at the request of the service subscriber.

The apparatus further includes a DTMF unit for receiving a dual tone multiple frequency (DTMF) signal for requesting text conversion from a call terminal of the service subscriber in a call, and the classification unit classifies the received DTMF signal.

The classification unit classifies voice data of origination and reception by referring to a source IP and a port of a SIP (Session Initiation Protocol) message, a reception source IP and a port, and a synchronization source ID of a Realtime Transfer Protocol (RTP) packet.

The apparatus may further include a silence remover configured to remove a silence indicator (SID) packet corresponding to a silence to leave only a voice packet among the classified RTP packets of the outgoing voice data and the received voice data, and the conversion unit may include the left voice packet. Convert to text using Speech To Text (STT) engine.

The apparatus further includes a timestamp unit for calculating a voice generation time as timestamp information of the converted text using the amount of time of the RTP packet referenced in the ptime value of the Session Description Protocol (SDP) of the SIP message.

The timestamp unit checks codec information of the RTP packet, confirms a timestamp that increases per second from the confirmed codec sampling rate, and calculates a time amount of the SID packet from a timestamp value of an SID packet increased from a previous packet. The voice generation time is calculated by using the time portion of the RTP packet and the calculated time portion of the SID packet.

The call text section stores the call text including at least one set of an outgoing call number, an incoming call number, total call time, outgoing text data and voice generation time, at least one set of received text data and voice generation time. do.

The device may be configured such that, at the start position of the first voice data received from the outgoing voice data and the received voice data, the voice data of the late arrival or reception of the late arrival or reception of the voice data arrives later by the time of the RTP packet of the first voice data. A mixing unit for setting a starting position and mixing each of the outgoing voice data and the received voice data streams into a single integrated voice data stream using the set start position, wherein the providing unit includes the integrated unit; Provide voice data.

The providing unit provides the call text to a terminal of a service subscriber using at least one of a text message, an email, a social network service (SNS), and a web page.

According to another aspect, a method in which a device text-converts a voice of a service subscriber and manages the converted text, comprises: receiving voice data of a call in which the service subscriber becomes a caller or a receiver; Classifying the received voice data into outgoing voice data and received voice data, respectively; Converting the classified voice data into data of an outgoing text and a received text, respectively; Generating a call text by dividing the outgoing text of the converted sender and the received text of the receiver and arranging them in chronological order; And querying the generated call text and providing the generated call text to a terminal of a service subscriber at the request of the service subscriber.

According to an aspect of the present invention, by managing the content of the call as text, the user can search and search the text of the call content as needed.

In text conversion of the call content, the processing load of the STT engine can be reduced by removing the silent packet and converting only the packet of the call voice.

When storing the caller's and receiver's conversations into a single piece of unified voice data, crosstalk is achieved by setting the starting position of the late arriving or receiving party's voice packet back by the amount of time of the first arriving or receiving party's packet. Can store one unified voice data removed.

The following drawings attached to this specification are illustrative of preferred embodiments of the present invention, and together with the detailed description of the invention to serve to further understand the technical spirit of the present invention, the present invention is a matter described in such drawings It should not be construed as limited to
1 is a schematic structural diagram of a call management system according to an embodiment of the present invention.
FIG. 2 is a schematic configuration diagram of the text management server of FIG. 1.
3 is an exemplary diagram in which the text management server of FIG. 2 classifies voice data of Tx and Rx.
4A and 4B are exemplary diagrams in which the text management server of FIG. 2 classifies call voice packets into outgoing voice packets and received voice packets.
5 is an exemplary diagram of an SID packet among call voice packets received by the text management server of FIG. 2.
6 is an exemplary diagram in which the text management server of FIG. 2 calculates a time stamp time of an SID packet of FIG. 5.
7 is an exemplary diagram in which the text management server of FIG. 1 mixes outgoing voice data and received voice data into one integrated voice stream according to another embodiment of the present invention.
8A and 8B are exemplary diagrams of delaying the start position of voice data of a transmitting or receiving side that arrives late in the mixing of FIG. 7.
9 is a schematic flowchart of a text management method of a call voice according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms or words used in the present specification and claims should not be construed as being limited to the ordinary or dictionary meanings, and the inventors should properly explain the concept of terms in order to best explain their own invention. It should be interpreted as meaning and concept corresponding to the technical idea of the present invention based on the principle that it can be defined. Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical spirit of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

1 is a schematic diagram of a call management system 100 according to an embodiment of the present invention.

Call management system 100 according to an embodiment of the present invention is a caller call terminal 110, a text management server 130, Telephony Application Server (TAS) 140, Call Session Control Function (CSCF) 150 and It is configured to include the receiver call terminal 170.

When the call call is sent from the caller call terminal 110 to the receiver call terminal 170, call data (call data) is relayed through the communication company server, and when the receiver call terminal 170 receives the incoming call, the call is connected. The caller's and receiver's voice call begins. The call data is divided into signal data (eg, SIP message) and voice data (eg, RTP packet), and each of the separated data has a separate processing and transmission path.

The carrier server in the VoLTE environment includes a TAS 140 and a CSCF 150 for processing the signal data. CSCF 150 relays the SIP message of call data from the call processing server. In addition, the TAS 140 is a server that provides a call-based multimedia supplementary service (eg, caller ID service, ring back tone service, call waiting service, etc.) to a subscriber of the supplementary service. When the call service subscriber subscribes to the call management service of the present invention, the service subscription information is managed in the TAS 140. The voice data of the call data is relayed to the caller call terminal 110 and the callee terminal 170 through the call processing server of the carrier.

Here, the present invention implements the call processing server as a text management server 130. The text management server 130 provides a text conversion service of call contents to the subscriber of the call management service in the process of relaying voice data of call data between the caller call terminal 110 and the callee terminal 170. That is, the call management service may convert the voice call contents into text at the request of the service subscriber and manage the converted text to provide the search and search service to the subscriber.

2 is a schematic diagram of the text management server 130 of FIG. 1.

Text management server 130 according to an embodiment of the present invention is a receiver 231, DTMF unit 232, classification unit 233, silence remover 234, converter 235, time stamp unit 236 ), A call text unit 237 and a provider 239.

Assuming that the text management server 130 is a computer terminal composed of a memory and a processor, each component 231 to 239 may be loaded into a memory in the form of a program and executed by the processor. For example, the components 231 to 239 are produced by a text management program, and then executed by a processor of the text management server 130 to provide the text management service using text converted from voice of a call content. can do.

The receiver 231 receives the voice data of the sender and the receiver generated during the call in an RTP packet. Of course, the outgoing voice data is relayed through the text management server 130 to the recipient call terminal 170 as a destination, and the received voice data is relayed through the text management server 130 to the caller call terminal 110 as a destination. do.

The DTMF unit 232 receives a DTMF signal for requesting text conversion of a voice call from a subscriber's call terminal during a subscriber's call subscribed to the call management service of the present invention. That is, the subscriber of the call management service generates a DTMF signal by pressing a specific key (eg, * key) when requesting text conversion of a call voice during a call.

When the DTMF signal of the text management service is confirmed by the DTMF unit 232, the classifier 233 classifies the voice data received by the receiver 231 into outgoing voice data and received voice data, respectively. For reference, classification of the voice data will be described later with reference to FIGS. 3 to 4B.

When the RTP packets of the outgoing voice data and the received voice data classified by the classifier 233 correspond to the silence, the silence remover 234 removes the silent RTP packets because the silence does not need to be converted into text. By this removal, only RTP packets containing call voice can be left. For reference, the silent RTP packet will be described later with reference to FIGS. 5 and 6.

The conversion unit 235 converts the voice of the outgoing RTP packet into outgoing text, and converts the voice of the incoming RTP packet into outgoing text. The conversion unit 235 may be configured as an STT engine and may be constructed as a separate STT server.

Here, the conversion unit 235 has an advantage that the data processing load is reduced because only the RTP packets of the call voice from which the silent RTP packet is removed by the silent remover 234 can be converted into STT-based text. .

The time stamp unit 236 calculates a time stamp value of the RTP packet as a time, and sets the calculated time as time stamp information corresponding to the talk time of the converted text. For example, the converted text of a call voice has a timestamp of the time of occurrence, like a message.

Here, the time stamp unit 236 calculates the time because the time of the text management server 130, the caller terminal 110 and the receiver call terminal 170 is different, and the path of the RTP packet and the path Because time is different. In addition, since the arrival order of the RTP packets arriving at the text management server 130 does not match the voice generation order, the arrival time should not be used as the voice generation time. In consideration of this point, the time stamp unit 236 sets the reference time based on the call start time, and calculates the occurrence time based on the value of the time stamp included in each RTP packet. Therefore, by the above calculation, the speech generation time of the text converted from the speech has the advantage of accuracy. The call start time corresponding to the reference time may be the arrival time of the first packet. For reference, the calculation will be described later with reference to FIG. 6.

The call text unit 237 generates a call text by dividing the caller's outgoing text and the caller's received text converted by the converter 235 and arranging them in chronological order calculated by the time stamp unit 356, and generating the call. Save the text.

Here, the stored call text information is provided to the service subscriber, so that at least one or more sets of the outgoing telephone number, the received telephone number, the total talk time, the outgoing text data and the voice generation time, the at least one of the received text data and the voice generation time It includes the above set. That is, the set consists of each text sentence and its speech generation time. The screen UI on which the call text information is displayed may be configured similarly to the messenger screen, and there is no particular limitation.

The provider 239 inquires the call text managed by the call text unit 237 at the request of the service subscriber and provides the call text to the terminal of the service subscriber.

The call text information may be provided to a terminal of a service subscriber using at least one of a text message, an email, a social network service (SNS), and a web page. For example, the web page in which caller and receiver call contents are listed as text information, such as a screen UI of katok, may be provided to a terminal of a service subscriber.

3 is an exemplary diagram in which the text management server 130 of FIG. 2 classifies voice data of Tx and Rx.

When the service subscriber A becomes the caller and calls the receiver, the Rx stream of the outgoing data is the voice content of A, and the Tx stream of the received data is the voice content of the receiver the caller A listens to.

The classification unit 233 of the text management server 130 may include an originating IP and port, a receiving IP, and a port included in the Session Initiation Protocol (SIP) message of the call data. The RTP packet of the call data is classified into a Tx stream of the outgoing voice data and an Rx stream of the received voice data by referring to a synchronized source ID (SSRC) of the RTP packet. SSRC corresponds to an identifier of a call source sound source connected to a stream by synchronization.

For example, if service subscriber A is a caller and calls a recipient, the RTP packets of source IP / Port is A's call terminal and destination IP / Port is text management server 130, indicating Classified as Tx stream. In addition, the RTP packets whose source IP / Port is the text management server 130 and the destination IP / Port is the call terminal of A are classified into Rx streams of the received voice data that the other party speaks.

4A and 4B are exemplary diagrams in which the text management server 130 of FIG. 2 classifies a call voice packet into an outgoing voice packet and a received voice packet.

Referring to FIG. 4A, a voice RTP packet of a sender and a receiver arriving at the text management server 130 is shown. Assuming that RTP packet 1 is the incoming voice and RTP packet 2 is the outgoing voice, three packets of RTP packet 1 on the receiving side first arrive at text management server 130 and then two packets of originating RTP packet 2 onwards. It arrives. As the conversation progresses during the call, each RTP packet arrives at the text management server 130.

Here, when the time portion of the RTP packet has 20 ms of voice data, it can be interpreted that the receiver of RTP packet 1 speaks for 60 ms first, and then the sender of RTP packet 2 speaks.

Referring to FIG. 4B, the classifier 233 classifies voice packets of the RTP packet 1 on the receiving side and the RTP packet 2 on the calling side with reference to the IP / Port value and the SSRC value as shown in FIG. 3.

5 is an exemplary diagram of an SID packet 500 among call voice packets received by the text management server of FIG. 2.

RTP packets of voice data are divided into voice packets containing real human voices and silent SID packets 500. The SID packet 500 is a silent RTP packet generated when the user stops talking and listens.

If you look at the actual conversation of the call, the other party hears while the caller or the receiver is talking. When voice data is separated into source Tx and receiver Rx data, a situation occurs in which approximately 1/2 of each of the separated data is silent. That is, one half of the RTP packets of the call data may be SID packets 500 corresponding to the silence.

If the value defined for each codec is checked in the "FT" field defined in the payload of the RTP packet, the silent SID packet 500 can be known. For reference, in the case of the AMR-NB codec of FIG. 5, when the value "8" of "FT" is confirmed, the corresponding RTP packet is the SID packet 500 in which the user's voice is silent. In the case of the AMR-WB codec, when the value "9" of "FT" is confirmed, it is the SID packet 500. For reference, a method of confirming codec information will be described later with reference to FIG. 6. The silence remover 234 filters the SID packet 500 by checking the value of the "FT" field in the RTP packet and leaves only the voice packet. If only the remaining voice packets are transmitted to the conversion unit 235 of the STT server and text converted, the load of the STT server is reduced.

6 is an exemplary diagram in which the text management server 130 of FIG. 2 calculates a time stamp time of the SID packet 500 of FIG. 5.

It is assumed that the n-1 th RTP packet, the n th RTP packet 600, the n + 1 th SID packet 500, and the n + 2 th RTP packet are listed according to the voice order. Each RTP packet contains the value of a timestamp. Packets containing voice in the RTP packet have the same amount of time, but since the amount of time of the SID packet 500 is different for each codec, the amount of time must be calculated from the information of each codec.

In the case of the RTP packet including the voice, the time stamp unit 236 of the text management server 130 defines the amount of time of the RTP packet including the voice as the ptime value (eg, 20 ms) of the SDP of the SIP message.

In the case of the RTP packet corresponding to the SID packet 500, the time stamp unit 236 calculates the amount of time with reference to the codec information. The time stamp unit 236 may check the codec header in the RTP payload or find the codec information from the length of the payload. For reference, the VoLTE terminal uses an EVS, AMR-WB, AMR-NB codec. For example, in the case of the EVS codec, the SID packet 500 is a silent packet and has a shorter payload length than the general voice packet, and the general voice packets are transmitted every 20 ms. The SID packet 500 is transmitted every 20 ms. It can be transmitted or delivered in longer periods (multiples of 20 ms). That is, the SID packet 500 is required to calculate the amount of time corresponding to the packet size, and the amount of time may be calculated using the timestamp value of the RTP packet.

For example, assuming that the time stamp of the n th RTP packet 600 is 1,000 and the time stamp of the n + 1 th SID packet 500 is 2,600, the time stamp unit 236 sets the difference value of the time stamp to 1,600. Calculate The timestamp field in the RTP packets 500 and 600 is a relative value related to the sampling rate of the call codec. Therefore, in order to calculate an absolute amount of time through the timestamp of the RTP, the sampling rate of the call codec must be known. The sampling rate can be known by checking the SDP of the SIP message when the call is established. Assuming that a call is established with the AMR-WB codec, the timestamp section 236 refers to the sampling rate 16 kHz of the AMR-WB codec from the SDP. Since 16 kHz means 16000 samplings per second, it is sampled 16 times in 1 ms, and the value of the time stamp of the RTP is increased by 16 per 1 ms. In the case of the voice RTP packet 600, since the RTP packet is transmitted every 20 ms, the value of the time stamp is increased by 320 (16 * 20).

Here, since the difference value of the time stamp of the RTP packet 600 and the SID packet 500 is calculated to be 1,600, 1,600 / 320 = 5, and 5 * 20 ms = 100 ms is calculated. Then, the time stamp unit 236 calculates the amount of time of the SID packet 500 to 100 ms. The combined amount of time for the RTP packet 600 and the SID packet 500 is calculated as 20 + 100 = 120 ms. Accordingly, in the order of the n-1 th RTP packet, the n th RTP packet 600, the n + 1 th SID packet 500, and the n + 2 th RTP packet, the time stamp unit 236 may be configured as the RTP packet 600. When time stamp information of time t1 is set in the text, the occurrence time t1 + 120 can be calculated in the text of the n + 2th RTP packet and set as time stamp information.

7 is an exemplary diagram in which the text management server 130 of FIG. 1 mixes outgoing voice data and received voice data into one integrated voice stream according to another embodiment of the present invention.

In another embodiment of the present invention, when the service subscriber receives the call text from the text management server 130 and confirms the call text, and requests the call voice content, the mixing unit of FIG. 2 provides the voice stream of the call content to the service subscriber. 238 mixes the outgoing call voice and the incoming call voice into one integrated voice.

Here, it is assumed that the RTP packet 1 stream 710 of the received voice and the RTP packet 2 stream 730 of the outgoing voice are mixed into one integrated voice stream.

As shown in FIG. 4B, when the start and the start positions of the RTP packets 400 and 410 are mixed in the same manner, cross talk occurs between the outgoing voice and the call voice. In order to prevent crosstalk and to mix the same voice stream as the actual content of the call as much as possible, the mixing unit 238 has three packets of the RTP packet 1 of the received voice first arriving at the RTP packet 2 stream 730 of the late-outed voice. Slows down by 3 * 20 ms = 60 ms.

8A and 8B are exemplary diagrams of delaying the start position of voice data of a transmitting or receiving side that arrives late in the mixing of FIG. 7.

Referring to FIG. 8A, actual call contents of A and B are shown based on the time axis. The call starts with A's voice followed by a call conversation between A and B.

Referring to FIG. 8B, when the start positions of the RTP packets of Tx and Rx are the same position or based on the arrival time of the packet without applying the technique of the present invention, in the regions 810 and 830 where the crosstalk is generated. The voices of B and B can overlap.

The present invention arranges and mixes the streams of the RTP packets of B at the start position, where the mixing unit 238 is delayed backward by the amount of time of the RTP packets of A, which are first arrived, to exclude cross talk. Thus, the call voices of A and B are mixed into one integrated call stream that excludes cross talk.

9 is a schematic flowchart of a text management method of a call voice according to an embodiment of the present invention.

When a call is initiated between the caller call terminal 110 and the callee terminal 170, the text management server 130 receives the call data of the origination and reception (S910). Here, the text management server 130 receives the SIP message of the signaling data through the TAS 140. The text management server also receives an RTP packet of voice data.

If the subscriber presses the key button of the call terminal to request the text conversion service of the present invention during the call, the text management server 130 receives the DTMF signal of the corresponding service key from the call terminal of the subscriber (S920).

By receiving the DTMF signal, the text management server 130 classifies the RTP packet of the outgoing voice data and the packet of the received voice data (S930). The classification of the packet is handled by reference to the originating IP and port, the receiving IP and port and the SSRC value as shown in FIG. In addition, as shown in Figure 4b, it is classified into a receiving side RTP packet 1 (400) and a sending side RTP packet 2 (410), respectively.

 When the RTP packets of the sender and the receiver are classified, the text management server 130 removes the silent SID packet 500 from each of the classified RTP packets and leaves RTP packets including voice (S940).

After the silent SID packet 500 is removed, the text conversion server 130 performs text conversion on the RTP packets including the voice using the STT engine (S950).

Here, the text conversion server 130 calculates the start time of each packet by setting the reference time as the initial call occurrence time with respect to each of the RTP packets 400 and 410 arranged in the voice generation order as shown in FIG. 4B (S951). ). As described above with reference to FIG. 6, the SID packet 500 is calculated with reference to codec information.

After the text conversion and time calculation are completed, the text conversion server 130 matches the converted test with the calculated speech generation time as one set, and includes a call including each matched set to be displayed in a preset screen UI. The text is generated and stored (S960).

Then, when the service subscriber requests the provision of the call text, the text conversion server 130 inquires and provides the call text corresponding to the request of the subscriber (S970). That is, the text conversion server 130 may provide a search and retrieval service of the call text changed from the call content according to the request of the service subscriber.

Although the present invention has been described by way of limited embodiments and drawings, the present invention is not limited thereto, and the technical spirit of the present invention and claims to be described below by those skilled in the art to which the present invention pertains. Of course, various modifications and variations are possible within the scope of the equivalents.

100: system 110: caller call terminal
120: text management server 140: TAS
150: CSCF 170: receiver call terminal

Claims (18)

An apparatus for text-converting call voice of a service subscriber and managing the converted text,
A receiver which receives voice data of a call in which the service subscriber becomes a caller or a receiver;
A classification unit which classifies the received voice data into outgoing voice data and received voice data, respectively;
A conversion unit for converting the classified voice data into data of an outgoing text and a received text, respectively;
A call text unit for generating a call text by dividing the outgoing text of the converted sender and the received text of the receiver and arranging them in chronological order; And
Providing unit for inquiring the generated call text and providing to the terminal of the service subscriber at the request of the service subscriber
Device comprising a. The method of claim 1,
And a DTMF unit for receiving a dual tone multiple frequency (DTMF) signal for requesting text conversion from a call terminal of the service subscriber in a call,
And the classification unit processes the classification by the received DTMF signal. The method of claim 1,
The classification unit,
And classifying voice data of origination and reception by referring to a source IP and a port of a SIP (Session Initiation Protocol) message, a synchronization source ID of a receiver IP and a port, and a Realtime Transfer Protocol (RTP) packet, respectively. The method of claim 1,
And a silence remover configured to remove silence indicator (SID) packets corresponding to silence to leave only voice packets among the classified RTP packets of the outgoing voice data and the received voice data.
And the converting unit converts the left speech packet into text using a speech to text (STT) engine. The method of claim 1,
And a timestamp unit for calculating a voice generation time as timestamp information of the converted text using the amount of time of the RTP packet referenced in the ptime value of the Session Description Protocol (SDP) of the SIP message. The method of claim 5,
The time stamp unit,
Confirming codec information of the RTP packet, checking a timestamp increasing per second from the confirmed codec sampling rate, calculating a time portion of the SID packet from the timestamp value of the SID packet increased from the previous packet, and checking the RTP The voice generation time is calculated using the time amount of the packet and the time amount of the calculated SID packet. The method of claim 1,
The call text unit,
Storing said call text comprising at least one set of an outgoing call number, an incoming call number, total talk time, outgoing text data and voice generation time, at least one set of received text data and voice generation time Device. The method of claim 1,
From the outgoing voice data and the received voice data, the start position of the voice data of the late arrival or reception of the late arrival or reception is set by delaying the time by the time of the RTP packet of the first arrived voice data. And a mixing unit configured to mix the voice streams of each of the outgoing voice data and the received voice data into one integrated voice data stream using the set start position.
And the providing unit provides the integrated voice data. The method of claim 1,
The providing unit,
And providing the call text to a terminal of a service subscriber using at least one of a text message, an email, a social network service, and a web page. A method for text-converting a voice of a service subscriber by a device and managing the converted text, the method comprising:
Receiving voice data of a call where the service subscriber becomes a caller or a receiver;
Classifying the received voice data into outgoing voice data and received voice data, respectively;
Converting the classified voice data into data of an outgoing text and a received text, respectively;
Generating a call text by dividing the outgoing text of the converted sender and the received text of the receiver and listing them in chronological order; And
At the request of the service subscriber, providing the generated call text to the terminal of the service subscriber
How to include. The method of claim 10,
Prior to the classifying step,
Receiving a dual tone multiple frequency (DTMF) signal for requesting text conversion from a call terminal of the service subscriber in a call;
And the classifying step is classifying the received DTMF signal. The method of claim 10,
The classifying step,
And classifying voice data of the originating and receiving by referring to the source IP and port of the SIP (Session Initiation Protocol) message, the source IP of the receiving side and the synchronization source ID of the Real Time Transfer Protocol (RTP) packet, respectively. Way. The method of claim 10,
After the classifying step,
Removing a silence indicator (SID) packet corresponding to a silence to leave only a voice packet among the classified RTP packets of the outgoing voice data and the received voice data,
The converting step is a step of converting the remaining speech packet to text using a Speech To Text (STT) engine. The method of claim 10,
After the converting step,
Calculating a voice generation time as time stamp information of the converted text using the amount of time of the RTP packet referred to in the ptime value of the Session Description Protocol (SDP) of the SIP message. The method of claim 10,
The calculating step,
Confirming codec information of the RTP packet, checking a timestamp increasing per second from the confirmed codec sampling rate, calculating a time portion of the SID packet from the timestamp value of the SID packet increased from the previous packet, and checking the RTP And calculating the voice generation time using the time amount of the packet and the time amount of the calculated SID packet. The method of claim 10,
The generating step,
Storing the call text comprising at least one set of an outgoing call number, an incoming call number, a total call time, outgoing text data and a timestamp, and at least one or more sets of incoming text data and a timestamp. Way. The method of claim 10,
Prior to the providing step,
From the outgoing voice data and the received voice data, the start position of the voice data of the late arrival or reception of the late arrival or reception is set by delaying the time by the time of the RTP packet of the first arrived voice data. And mixing the voice streams of each of the outgoing voice data and the received voice data into one integrated voice data stream using the set start position.
The providing step is providing the integrated voice data. The method of claim 10,
The providing step,
And providing the call text to a terminal of a service subscriber using at least one of a text message, an email, a social network service, and a web page.

Images