KR19990015749A - Recognize your voice and email - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a technique for converting text input into text information and transmitting the same, and more particularly, to an apparatus and a method for sending an email by recognizing a voice.

According to the conventional e-mail processing apparatus, there is a limitation that an e-mail content can be created and transmitted only in a system such as a PC or a workstation where an e-mail server or a client program is installed.

According to the present invention, when a user inputs a content corresponding to an e-mail to be transmitted as a voice signal according to a series of promised procedures, the user recognizes the voice, creates a mail of the corresponding content, and transmits the created mail to a recipient system. The present invention provides an apparatus and method for sending an e-mail by recognizing a voice.

In particular, the present invention by inputting the contents of the e-mail by voice using a telephone line even in a remote place without an e-mail server or a client program is installed, and recognizes the voice to be able to transmit the e-mail Provided are a device and method for sending e-mail.

Description

Apparatus and method for sending e-mails by recognizing voices

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a technique for converting text input into text information and transmitting the same, and more particularly, to an apparatus and a method for sending an email by recognizing a voice.

1 is a block diagram of a conventional e-mail processing apparatus, which is provided in a PC or other terminal including PC communication related components.

The communication connection unit 101 is a circuit for connecting to a server using a telephone line as a modem or a LAN card, and for receiving data from the server and providing the data to the terminal. The microcomputer unit 102 connected to the communication connection unit 101 is e. Control related to the transmission of e-mail, that is, the e-mail is transmitted and controlled by performing communication with a server through a communication connection unit, and the memory unit 103 is connected to the microcomputer 102 and received from the server. The e-mail is stored, and the video unit 104 is connected to the microcomputer 102 to convert the received e-mail into an analog signal, and the display unit 105 outputs the e-mail output from the video unit 104. -mail is displayed as a video signal on the screen.

According to the conventional e-mail processing apparatus shown in FIG. 1, there is a limitation that an e-mail content can be created and transmitted only in a system such as a PC or a workstation in which an e-mail server or a client program is installed.

According to the present invention, when a user inputs a content corresponding to an e-mail to be transmitted as a voice signal according to a series of promised procedures, the user recognizes the voice, creates a mail of the corresponding content, and transmits the created mail to a recipient system. The present invention provides an apparatus and method for sending an e-mail by recognizing a voice.

In particular, the present invention by inputting the contents of the e-mail by voice using a telephone line even in a remote place without an e-mail server or a client program is installed, and recognizes the voice to be able to transmit the e-mail Provided are a device and method for sending e-mail.

1 is a block diagram of a conventional e-mail processing apparatus

2 is a diagram illustrating the concept of transmitting an e-mail of the present invention by voice.

Figure 3 is a block diagram of an e-mail transmission device of the present invention

4 is a flowchart of a voice recognition pattern generation process in the present invention

Figure 5 is a voice signal processing waveform diagram for generating a voice recognition pattern in the present invention

6 illustrates an e-mail voice data format in the present invention.

Figure 7a, 7b is a flow chart of the process of transmitting e-mail by voice in the present invention

FIG. 8 is a flowchart showing subroutine A of FIG. 7 in detail.

9 is a flowchart showing in detail the subroutine B of FIG.

2 is a diagram illustrating a concept of a system for sending an e-mail by recognizing a voice of the present invention, wherein a user inputs a voice signal using a telephone or a microphone and recognizes the input voice in a user system. A text message corresponding to the recognized voice content is created. The e-mail data thus created is connected to an e-mail server for transmission, and the e-mail server is connected to a recipient system to receive and confirm the voice / mail converted e-mail.

The created email data shows a textual email document (file) with the recipient's email address, subject, and body.

Fig. 3 is a block diagram of a device for sending an e-mail by recognizing the voice of the present invention, which establishes a communication line between a telephone network and a system and transmits and receives e-mail information. And a voice processing unit 202 for converting a voice signal input to the communication connection unit and an external microphone into digital data, and controlling the communication connection unit 201 to transmit and receive e-mail information and recognize a voice signal. A microcomputer 203 for creating a mail, and a memory unit for storing e-mail messages generated by recognizing voice data, an address book of a user, and the like, under the control of the microcomputer 203, and storing voice data of the voice processing unit. 204 and a video unit 205 for displaying the e-mail data transmitted and received under the control of the microcomputer 203 on the display unit 206.

The communication connection unit 201 under the control of the microcomputer 203 is connected to a telephone line or a LAN line, a switching unit for switching the telephone line or a LAN line, a tone generator for generating a dial tone, and a call for detecting an incoming call. Phone number or IP under the control of the microcomputer 203, including a tone detector, a digital analog converter for data transmission and reception, an analog digital converter and a communication modem, a controller for controlling them appropriately, and an interface with the microcomputer 203. The communication line is established by the address.

The voice processing unit 202 includes an A / D converter for converting an input analog voice signal into a digital signal, a filter for removing noise of the voice signal, or filtering for converting to a frequency band suitable for a voice transmission band of a telephone network. Circuits and the like.

The microcomputer 203 includes means for performing a known speech recognition algorithm for speech recognition, and means for creating the recognized speech into a file in an e-mail format.

The memory unit 204 includes digitally converted voice data, an e-mail message (data) generated by the microcomputer 203 by recognizing the voice data, for example, an euc-kr code and voice pattern data associated therewith, Information such as the user's address book (recipient's name and recipient e-mail address associated with it) is stored.

As shown in Fig. 2 and Fig. 3, a user who makes a phone call from the outside and recognizes the voice of the present invention through the communication connection unit 201 and sends an e-mail or a user equipped with this function Upon connecting to the system, the microcomputer 203 receives the voice signal input by the user as digital data through the voice processing unit 202, recognizes the user's voice from the pre-stored user's voice recognition pattern information, and indicates the recognized voice. Create an e-mail and send it to the recipient system.

At this time, the recognized voice data and the created e-mail message are stored in the memory unit 204. On the other hand, when the user requests, the displayed e-mail is displayed on the display unit 206 through the video unit 205. .

Figure 4 is a process for generating a pattern for voice recognition of the user in the present invention described above, Figure 5 shows an end point detection section as an example of the speech recognition pattern, Figure 6 is e as a result of speech recognition -mail is a voice data format, and FIGS. 7 to 9 show a process of creating and transmitting an e-mail.

A voice recognition pattern generation process of the user by the voice processor 202 and the microcomputer 201 will be described with reference to FIGS. 4 and 5.

The pattern generation method used as a reference for speech recognition uses well-known techniques.

'Pattern Matching Speech Recognition Method' is a technology that recognizes speech by comparing inputted voices with voice patterns made by the same method by using a method such as frequency conversion and 'Neural Network Speech Recognition Method'. Is a method of automatically recognizing input voice using patterns for neural network learning.

In the present invention, the voice recognition technology, which is widely provided in software, is applied, and when there are several users, since the characteristics of the voice are different for each person, the recognition rate is lowered and the recognition rate is higher for a specific person. In view of the above, the present invention enables a high recognition rate on the premise of recognizing one user in one service system.

In addition, not only the voice recognition pattern based on the user's own voice pattern, but also a method of recognizing voices of various users by constructing a standard voice of a prominent announcer as a recognition database.

In FIG. 4, the service system of the present invention, which recognizes a voice and sends an e-mail, allows the user to make a voice pattern for the entire Hangul code by using a voice processing unit through a microphone for each of the Hangul codes. 202) to input voice.

When the voice according to the complete Hangul code is input, the voice processor 202 converts the voice into an analog signal through appropriate filtering, and notifies the start and end of the input voice using a specific key.

This digitally converted voice data is subjected to an end point detection process as shown in FIG.

The end point detection process is a technology that actually detects only the speech segment from the stored speech data. By detecting the change of amplitude on the time axis, the speech segment is distinguished from the speech segment and the non-voice segment. The voice pattern is stored in the memory unit 204.

After this process is performed for all Korean codes, a voice pattern for a space is generated.

That is, if a voice is input for 1 to 2 seconds without inputting a micro voice and detecting the end point, it is difficult to separate a section because there is no amplitude above a certain level, and thus the voice data stored at the center of the stored voice data section 3/4 voice sections are randomly picked and stored as voice patterns.

By using this, even if there is no voice among the voice data input through the actual telephone or a signal due to noise component, etc., there is no matching pattern so that the space can be processed.

Through the above process, all the voice patterns of the user are combined with the Hangul code and stored.

Hangul code is complete and uses euc-kr named according to the method of naming the character codes commonly used on the Internet.

Finally, the memory unit 204 combines the recipient's name and the recipient's e-mail address to create and store an address book.

The contents stored in the memory unit 204 are well represented in the memory unit 204 of FIG.

After the process of generating and storing the voice recognition pattern, the user recognizes the voice input by using the telephone at a distance using the voice recognition pattern, and creates and directs an e-mail as the recognized voice indicates. 7 to 9 are transmitted to the receiver system.

That is, the voice input through the telephone network is stored in the memory unit 204 in the form of an e-mail voice data format as shown in Fig. 6 by the voice processing unit 202, and this e-mail voice data is stored as e as shown in Fig. 2. It is written as -mail data and is sent to the recipient.

As can be seen from Fig. 6, the e-mail voice data has a byte or word size of the data before each data, and the ASCII value and the title voice of the receiver previously recognized by the system. It consists of the number of data, the subject voice data, the number of body voice data, and the body voice data which is the actual content of the mail, and also the carriage return information (line feed) by recognizing dial tone (for example #). The value specified by is stored in the body text.

As shown in Fig. 2, the e-mail data generated as a result of the voice recognition is formed of an euc-kr code, and the e-mail address of the recipient is stored after the To code for use in e-mail transmission. Next, the subject of the e-mail and the body of the e-mail are stored after the body.

When transmitting this to the receiver system, the service system of the present invention transmits the mail by adding other additional e-mail elements, for example, the sender, the date, and the Hangul coding information, using the created e-mail.

See FIG. 7.

The microcomputer 203 controls the communication connection unit 201 and uses the auto answering function to check the authenticity of the user when the user dials his system and inputs a promised key (password) for sending an e-mail. Activates the function to send mail.

If the key entered by the user is justified, an OK tone is sent and the voice data format for e-mail creation is executed according to the promised dial tone.

As promised dial tone, '*' key ends the voice, '#' key wraps, '1' key recipients, '2' key headings, '3' key body, and '0' key ends It is defined as

When the '1' key is input through the communication connection unit 201, the microcomputer 203 recognizes this and processes the input voice as a 'receiver' until the * key is input after transmitting the OK tone (digital conversion and temporarily storing). do.

Alternatively, a specific recipient may be mapped to a specific key number, and a recipient name corresponding to the key number may be searched by inputting the promised key number instead of recognizing the recipient name by voice.

When '2' key is input, the input voice is treated as 'title' until * key is input.

That is, when the '2' key is input, the voice signal pronounced the name of the 'recipient' stored in the memory is recognized once, and it is determined whether the recognized receiver exists in the address book of the memory unit 204. Send an (Error) tone to wait for re-entry, if present, store it in the e-mail voice data with the recipient (ASCII code) and the receiver data size, then store the received voice data with the voice.

When the '3' key is input, the voice signal input afterwards is recognized as the body content of the mail, and the number of previously stored titles is stored in the memory, and the input voices are stored in sequence with the size.

After that, the text is recognized and stored until the '0' key is entered, and CRLF is executed when the # key is entered.

When the '0' key is input, the contents of the mail body are recognized as being finished, the number of the received voices is calculated and stored, the OK tone is transmitted, the stored voice data is recognized and the e-mail is created. Perform the process of transmitting it.

This process is illustrated in detail in FIGS. 8 and 9.

The microcomputer 203 reads the data size, which is the first part of the voice data, and the actual data 'receiver' code loaded thereafter, finds the corresponding e-mail address by referring to the address book, and then e with the 'To' code. Store in mail data.

In the next step, the title voice data stored as voice is loaded using the number of voices of the title and the respective voice size information. Then, the voice recognition is performed by comparing the recognition patterns stored in the memory unit 204. The subject of the recognized voice is converted into an euc-kr code and stored with the 'Subject' code in the e-mail file.

The recognition process is shown in FIG.

The routine for recognizing the subject voice uses the same subroutine (B) as for recognizing the body content, so that the next information to be read may be the size of the voice data or the carriage return information (CRLF).

In the case of the title, since the size of the voice data is read, the voice data is loaded according to the size information, and the end point detection for detecting the actual voice section is executed, and the pre-stored voice recognition pattern information is used for the end point detection process. After recognizing, euc-kr code is stored in e-mail.

In this case, when there is no voice, a code corresponding to a space is stored in an e-mail.

In this way, after recognizing the number of voice data of the title and storing the code, the routine returns to the routine of FIG. 8 to read, recognize and write the content of the text.

Before reading the text, save the 'Body' code in the e-mail data so that it can be distinguished when sending the e-mail, and read the number of voices in the text in the same procedure as in the routine of FIG. Load it and recognize it.

In this case, if the # key is used instead of the data size, the carriage return-line transfer information is stored in the carriage return code in the e-mail data.

The reason for not using the original carriage return code when storing the # key coming from the telephone line in the middle of voice data is that the system can recognize this code as the data size.

In this way, all voice data becomes e-mail data created as shown in FIG.

The next action is to send the mail to the recipient. The system uses the information in the e-mail data to create the actual e-mail file.

This is to use the receiver, the subject, the body content as it is, add the date, sender, mail coding type, etc., and then send the mail by accessing the e-mail server through the communication connection unit 201.

Further, in a service system for sending an e-mail by recognizing the voice of the present invention, an application for processing the recognized voice data into a format suitable for the facsimile transmission standard, creating a facsimile file, and automatically transmitting the same to a designated destination. Is also easy.

More broadly, it is also easy to apply an application to edit a voice-recognized data into a file complying with a predetermined transmission standard, and to automatically transmit it to a designated destination.

According to the present invention, it is possible to make and send a mail by telephone voice even from a remote site, and if the user ID is required, the e-mail can be transmitted regardless of the place and the device.

Claims (11)

Means for recognizing a voice signal input by a user in accordance with a predetermined transmission standard, means for creating an e-mail data corresponding to a predetermined transmission standard using the recognized voice signal, and the generated e-mail data. A device for sending an e-mail by recognizing a voice comprising a means for automatically transmitting to a recognized destination. A first step of checking a user's e-mail transmission request using a telephone line, a second step of receiving a user's voice whose authenticity is determined in the first step, and recognizing the voice, and a voice recognized at the second step The third step of creating an e-mail by processing the text information corresponding to the e-mail transmission standard and the fourth step of automatically transmitting the e-mail created in the third step to a recognized destination. How to send an e-mail by recognizing a voice characterized in that. According to claim 2, In order to meet the e-mail transmission standard in the voice input, the phone recognizes the e-mail by recognizing the voice, characterized in that the key processing information characterized in that the process is processed. How to send. 3. The method of claim 2, wherein the recognition of the destination is recognized by a predetermined key input. The method of claim 2, wherein the voice recognition e-mail service system is a user system. The method of claim 2, wherein the voice recognition e-mail service system is an e-mail server. The method of claim 2, wherein the first step of confirming the transmission request comprises: a; A method of sending an e-mail by recognizing a voice, characterized in that it comprises a process of checking a transmission request and a process of determining the authenticity of the user. The method of claim 2, wherein the second process of recognizing the voice comprises; A voice is inputted through a current microphone or a telephone, a process of comparing the input signal with a pre-generated pattern, and a process of recognizing a specific pattern based on a comparison result and storing the result. How to recognize and send e-mail. The method of claim 2, wherein the fourth process of automatic transmission comprises; A recipient recognition process of recognizing the recipient by reading the recipient memory data stored in the storage means and referring to the address book; and automatically transmitting a mail to the recognized recipient. how to send mail). The method of claim 2, further comprising a method of creating a space pattern by inputting unvoiced sound for a predetermined time and grabbing about 3/4 of the voice section of the stored voice data section at the center of the stored voice data section. How to send an e-mail by recognizing a voice. The method of claim 8, wherein the pattern generation process; Inputting voices for all the Hangul codes of the user, End-Point Detection process for detecting the end point by the change of amplitude, and Generation pattern for storing the frequency components of the detected signals in association with the Hangul code A method of sending an e-mail by recognizing a voice comprising a storing process.