KR0175251B1 - Automated response system using phoneme conversion method - Google Patents

Abstract

In order to change the service contents in the automatic response system, the voice must be newly recorded, the voice must be compressed digitally, and a large amount of storage device is required. Improve what is impossible to do.

To this end, a host computer for storing files of text data according to service contents, a line interface unit for interfacing with a public switched telephone network, and digitally encoded audio data are outputted to the line interface unit and inputted through the line interface unit. A codec for digitally encoding a speech signal, a speech processing means for analyzing the digitally encoded speech data, converting synthesized speech data into a digitally encoded speech data, and applying the same to the codec; A control means for receiving the text data of the corresponding service information from the host computer in response to the service request of the user according to the analysis of the means, converting the text data into synthesized sound data and applying it to the voice processing means; and a service execution program of the control means. And during operation And a first memory means for temporarily storing the generated data, a second memory means for storing a phoneme conversion control program of the control means, a voice database for the phoneme conversion, and an electronic dictionary.

Therefore, the recording of the contents of the service is completed only by creating a text file, and there is no need for a separate auxiliary memory device, and the information service is performed with high quality natural voice.

Description

Automated Response System Using Phoneme Conversion

1 is a block diagram of a conventional automatic answering system.

2 is a block diagram of an automatic response system according to the present invention.

3 is a flowchart of an auto answer function according to the present invention.

4 is a block diagram of a translator system for carrying out the present invention.

* Explanation of symbols for main parts of the drawings

10: host computer 12: control unit

22,28: 1st, 2nd memory part 30: Transfer buffer part

36: voice processor 44: codec

46: line interface unit

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic answering system (ARS), and more particularly, to an automatic answering system that converts text data of a text file into speech data by a phonetic conversion method and sends a voice message to a user. It is about.

In general, an automatic answering system includes a micro processing unit (MPU) 1, a main memory 2, an auxiliary memory 3, a codec 4 and a line as shown in FIG. It consists of a line interface part 5. In the automatic answering system of FIG. 1, the analog voice message input from the public switching telephone network connected to the line interface unit 5 is controlled by the MPU 1 under the control of the ADPCM (codec 4). Voice compression coding (Adaptive Differential Pulse Code Modulation) method to store in the secondary storage device (3), or in response to the user's request through the public switched telephone network by the control of the MPU (1) The ADPCM voice data stored in the RXM is read out, converted into an analog voice signal by the codec 4, and transmitted to the public switched telephone network.

In this case, the operation of transmitting a voice message in response to a user's request will be described in detail as follows. First, when the user calls the telephone system of the answering machine of FIG. 1, the line interface unit 5 detects an incoming ring signal and informs the MPU 1. The MPU 1 then turns on a relay in the line interface 5 in the call waiting state to form a call path, and then stores the voice data of the guide message for use of the auto answering system. Read from and output to codec (4). Accordingly, the guide message is sent to the user. When the user who listens to the guidance message presses the desired service number, the line interface unit 5 detects the number digit of the dual tone multi-frequency (DTMF) signal and informs the MPU 1 of the number. The MPU 1 then checks the service number and reads out the corresponding voice data from the auxiliary memory 3 and outputs it to the codec 4. Accordingly, the voice message requested by the user is sent to the user. When the user who listens to the voice message of the service hangs up, the MPU 1 detects a busy tone through the line interface unit 5, turns off the relay, restores the call path, and then returns to the call waiting state. do.

As described above, in order to change the service contents, the conventional automatic response system records a new voice on an analog tape, converts the recorded voice into digital (usually using the ADPCM method), and changes or adds the service voice to the auxiliary memory. Should be. In addition, since the compression efficiency of the voice compression by the ADPCM method is low during recording, a large capacity memory device is required, and it is impossible to realize natural sound by using the sound synthesis method by simple combining during playback.

Therefore, the present invention provides an automatic answering system that can provide a user with a high quality voice message that is closer to natural voice by translating a sound by simply registering the text data according to the service content as a text file without recording the voice message of the service content. In providing.

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

2 is a block diagram of an automatic answering system according to the present invention, which is a line interface for interfacing with a host computer 10 storing text files of text data according to service contents and a public telephone switching network under predetermined control. A codec 44 for digitally converting the digitally encoded voice data and outputting the analog coded audio data to the line interface unit 46 and digitally encoding the voice signal input through the line interface unit, and the digital codec 44. The voice processing unit 36 and the line interface unit 46 are controlled by converting the synthesized voice data into digital coded voice data and applying the encoded voice data to the codec 44. In response to the service request of the user according to the analysis of the voice processor 36, and recalls text data of the corresponding service information. Transmission for inputting and outputting data between the control unit 12 and the voice processing unit 36 and the control unit 12 which are received from the host computer 10 and converted into synthesized sound data and applied to the voice processing unit 36. A buffer unit 30, a first memory unit 22 for temporarily storing a service execution program of the control unit 12 and data generated during operation, and a phoneme conversion control program of the control unit 12 A second memory unit 28 for storing a voice database and an electronic dictionary, and an amplifier 48 for amplifying and applying a voice signal input and output between the codec 44 and the line interface unit 46 to the speaker 50; It consists of.

In the configuration of FIG. 2, the control unit 12 includes a central processing unit (CPU) 14, a serial input / output interface unit 16, a parallel input / output interface unit 18, and a timer 20. The first memory unit 22 is composed of an SRAM (Static RAM) 24 and a PROM (Programmable ROM) 26. The transmission buffer unit 30 is composed of a reception FIFO (First Input-First Output) and a transmission FIFO 34. The voice processor 36 includes a digital signal processor (DSP) 38, a PROM 40, and a serial input / output interface unit 42.

3 is a flowchart of an auto answer function according to the present invention.

4 is a block diagram of a translator system for carrying out the present invention.

Hereinafter, an operation of an embodiment of the present invention will be described in detail with reference to FIGS. 2 to 3.

First, in FIG. 2, the CPU 14 of the control unit 12 uses a CPU that has a 16-bit data bus 102, operates by a 16 MHz clock signal, and can address a memory of at least 8 M bytes or more. The serial input / output interface unit 16 is used for connection with the host computer 10. The parallel I / O interface unit 18 detects the ring state of the telephone line of the public switched telephone network and controls the relay on / off. The timer 20 generates an interrupt every predetermined time to check the state of the telephone line so as to process the DTMF signal and the call progress tone.

The host computer 10 is responsible for providing information and transmits text data of service contents desired by the user to the control unit 12 through the RS-232C 101.

The SRAM 24 of the first memory unit 22 stores addresses of various interrupt service programs processed by the CPU 14, and stores temporary data generated in the course of executing the monitor program and the translator control program.

The PROM 26 stores a system bug program and a monitor program of the CPU 14, and has a size of 512K bytes. The monitor program also handles system-wide data and processes events that occur on the line. The second memory unit 4 stores a phoneme conversion control program, a voice database, and an electronic dictionary in Korean and English, and has a size of 4M bytes (word conversion control program: 500k bytes, voice database: 1.2M bytes, electronic dictionary 1.8). M bytes). Therefore, the control unit 12 receives synthesized sound data from the monitor to generate synthesized sound data capable of generating a synthesized sound.

At this time, an electronic dictionary is required to interpret the document, and the synthesized sound data is generated by combining each component of the voice database. The speech database is composed of 1728 kinds of synthesized basic units, each of which is compressed to 10.5 Kbps by LSP analysis. Using this voice database, unlimited voice synthesis is possible with male and female voices.

The reception FIFO 32 of the transmission buffer unit 30 is a buffer for receiving the DTMF signal and the call tone information from the DSP 38, and the transmission FIFO 34 transmits the synthesized sound data to the DSP 38. Is a buffer.

The PROM 40 of the speech processor 36 stores an LSP speech synthesizer, a DTMF signal, and a call progress tone detection program, and has a size of 8K bytes. The DSP 38 is used to implement an LSP synthesizer that requires high-speed computation for real-time processing. The DSP 38 receives the synthesized sound data from the control unit 12 and sends the μ-law PCM data to the serial input / output interface unit 42 at a period of 8 ms. Output The serial input / output interface unit 42 inputs the DTMF signal of the telephone line and the call progress tone, interprets each signal, and informs the control unit 12 of the result.

The codec 44 converts the DTMF signal of the telephone line and the call tone to μ-law PCM data and transmits it to the control unit 12. The codec 44 receives the synthesized sound data through the serial input / output interface unit 42 and converts the analog voice into an analog voice line. Transfer to the interface unit 46.

The line interface unit 46 detects the ring signal of the telephone line, notifies the parallel input / output interface unit 18 of the control unit 12 to the ring state, and performs on / off operation of the relay under the control of the control unit 12. .

Meanwhile, in order to monitor the call state of the telephone line, the guidance message may be listened to through the amplifier 48 and the speaker 50.

Referring to FIG. 3, an operation of performing an automatic response in response to a user's call by the automatic answering apparatus of the present invention as described above is as follows.

First, when the user calls the telephone system of the answering machine of FIG. 2, the line interface unit 46 detects an incoming ring signal and notifies the control unit 12. FIG. Then, the control unit 12 forms a call path by turning on the relay of the line interface unit 46 in steps 201 to 203 and transmits the text data of the guide message for use of the automatic answering system to the host computer 10. After receiving and receiving the request, the sound-tone conversion is performed, the synthesized sound data is applied to the speech processing unit 36, and the speech processing unit 36 is controlled to output to the codec 44 as PCM data, which is digitally encoded speech data. Accordingly, the guide message is sent to the user. The user hears the guide message and presses the desired service number. Accordingly, the DTMF signal is input to the voice processing unit 36 through the line interface unit 46 and the codec 44. The sound processor 36 detects the number digits of the DTMF signal and informs the controller 12. Then, the control unit 12 checks the service number in steps 204 to 207, requests the host computer 10 to receive the text data of the corresponding voice message, and then converts the sound to synthesize the synthesized sound data. The audio processing unit 36 controls the speech processing unit 36 to output the coded data as PCM data, which is digitally encoded speech data. Accordingly, the voice message requested by the user is sent to the user. When the user who listens to the voice message of the service hangs up, the control unit 12 detects a busy tone through the line interface unit 46, the codec 44, and the voice processing unit 36 in steps 208 to 209. The relay will be turned off to restore the call and return to the call-waiting state.

In this case, the sound conversion is performed as shown in FIG.

As described above, the present invention has the advantage that it can be used as an information guide system that can be easily and easily changed every time by completing the recording of the service contents only by creating a text file without recording the voice message of the service contents. In addition, by constructing a voice database that enables unlimited voice synthesis with only a small amount of memory, it does not require a separate auxiliary memory device. Since it is not a simple sound synthesis method, it can provide information service with high quality natural voice close to human speech. Not only can the same announcement message be serviced by changing the voice of male and female, but also the service can be changed by changing the high, low, long and short speed of the service voice.

Claims (3)

In the automatic answering system using the phoneme conversion method, an analog conversion is performed between a host computer storing a file of character data according to service contents, a line interface unit for interfacing with a public switched telephone network under predetermined control, and digitally encoded voice data. A codec for digitally encoding a voice signal input through the line interface unit and a digital coded voice data of the codec; Control the line interface unit and the line interface unit to answer a call from the outside, and receive text data of the corresponding service information from the host computer in response to a user's service request according to the analysis of the voice processing unit Synthetic sound day A control means for converting the data into the voice processing means, a first memory means for temporarily storing a service execution program of the control means, and data generated during the operation; And a second memory means for storing a voice database and an electronic dictionary. The method of claim 1, further comprising a receiving buffer for applying the analysis data of the voice processing means to the control means, and a transmission buffer for applying the synthesized sound data of the control means to the voice processing means. Automatic response system using the phoneme conversion method. The automatic response system using a phonetic conversion method according to claim 1, further comprising a voice output means for converting and outputting a voice signal input and output between the codec and the line interface unit into voice.