JP2000022830A - Telephone sound recognition/response device, telephone sound recognition system and recording medium recording telephone sound recognition system control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To constitute a device to be compact and to prevent a whole system from becoming large even if the number of lines to be recognized increases. SOLUTION: A sound recognition/response device has a CPU board 21, a telephone response board 22, an A/D conversion board 23, an HDD device 25, and a sound synthesis device 26. A sound signal from a telephone line is inputted to the A/D conversion board 23 via the telephone response board 22. The A/D conversion board 23 converts the inputted sound signal into a digital signal, records it and transmits the recorded sound signal to the CPU board 21 through an inner bus 24. The CPU board 21 recognizes the transmitted sound signal and takes out response information stored in the HDD device 25 based on the recognition processing. Response information is sent to the sound synthesis device 26 and converted into the synthesized sound signal. The converted sound signal is transmitted to the telephone line via the telephone response board 22.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
The present invention relates to a response device, a telephone voice recognition system, and a recording medium storing a telephone voice recognition system control program.

[0002]

2. Description of the Related Art As a speech recognition apparatus, there is a discrete word speech recognition system shown in FIG. In this system, as shown in FIG. 10, voice data is input to a voice input unit 12 from a voice input device 11 such as a telephone or a microphone. The audio data input to the audio input unit 12 is supplied to a feature extraction unit 13, where the audio data is subjected to frequency analysis. The data obtains a spectrum sequence from the result of the frequency analysis and is input to the phoneme recognition unit 14. Phoneme recognition unit 14
Is composed of a neural network whose output is duplicated.

The neural network of the phoneme recognition unit 14 includes an input layer, a hidden layer, and an output layer. For example, a spectrum of five frames is input to the input layer every time. The central spectrum sends the corresponding phoneme according to the value of the unit in the output layer. Since the output units are duplicated, two units are associated with each phoneme category. On the other hand, as a result, two units are selected from those having the largest output value, and the phonemes corresponding to the two units are obtained as the first and second phoneme candidates.

[0004] The similarity between the recognized phoneme candidate sequence and the template 15 in the dictionary having the phoneme pattern of the vocabulary to be recognized is determined by the first place in the phoneme candidate sequence recognized as the phoneme in the template. And the degree of similarity with the second candidate is defined as a local score. The local score is matched by the matching unit 16 by the DTW method or the like, and the overall similarity score is obtained by accumulating the most similar words. The word having the minimum similarity score among all the vocabularies to be recognized is output from the matching unit 16 as a recognition result.

[0005]

(1) There is a system in which the above speech recognition device is applied to a telephone speech recognition device. In this system, when a plurality of telephone speech recognition devices are operated simultaneously, it is common to individually process the results of each speech recognition device. When processing is performed individually in this way, the output results from other speech recognition devices are:
Since they cannot be referred to each other, there is a problem that each can perform only independent processing. Further, when it is necessary to process each voice recognition device in an integrated manner, it is desired to construct a client-server system that can process the results of all voice recognition devices collectively by one server. .

(2) In a server-client system having a telephone speech recognition device as a client, usability changes depending on how the control of each client (such as a shutdown process) can be performed from the server.
In order not to put an extra burden on the user, it is desired to construct a system that can easily control the client.

(3) Pattern matching including speech recognition processing requires a large amount of calculation, and in many cases, the entire computer must be used exclusively for recognition at the current processing speed of a computer. Therefore, even if the number of telephone lines used for recognition is increased by one, one computer must be newly added, and there is a problem that the scale of the system increases. As a result, in the case of configuring a system that can perform recognition processing simultaneously on a plurality of telephone lines, it is desired to manufacture a compact speech recognition device.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and it is possible to make the apparatus compact, to prevent the entire system from being enlarged even when the number of lines to be recognized increases, and to make each recognition client It is an object of the present invention to provide a telephone voice recognition / response device, a telephone voice recognition system, and a recording medium which records a telephone voice recognition system control program, which can collectively process the above processing results and can easily control.

[0009]

SUMMARY OF THE INVENTION In order to achieve the above-mentioned object, the present invention provides a CPU board having various memories mounted thereon and an A / D conversion board for converting an analog signal into a digital signal. A telephone answering board on which a network terminating device and a digital switch are mounted and a telephone line connection terminal is provided, a conductor for electrically connecting these boards, and a signal output from the CPU board are synthesized voice signals. A / D conversion of an analog signal sent from a telephone line connected to a telephone line connection terminal of the telephone answering board from the telephone answering board. A transmitted to the board
The digital signal is converted to a digital signal by a / D conversion board, the digital signal is transmitted to a CPU board, recognition processing is performed, and the recognition result is converted into a synthesized voice by a voice synthesizer, and then transmitted from the telephone answering board to the telephone line. It is something to do.

A second invention is to connect a telephone voice recognition / response device as a client to a plurality of networks and to transmit a recognition result to a server via the network.
The client receives a command from the server and automatically performs a process corresponding to the command.

[0011] A third invention is a recording medium recording a program for controlling a telephone voice recognition system, comprising means for using a status notification function provided by a server, not only a status notification to the server but also a server. From the server to the client via a return value, and a program functioning as a means for controlling the client from the server.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is an explanatory diagram of a configuration of a voice recognition / response device according to a first embodiment of the present invention. In FIG. 1, reference numeral 21 denotes a CPU board, and in addition to the CPU, the CPU board 21 includes a DRAM, a DPRAM, and a network. A card, a video card, and the like are provided. Reference numeral 22 denotes a telephone answering board.
It is configured as shown in FIG. An A / D 23 receives an audio signal from a telephone line via the telephone answering board 22.
The A / D conversion board 23 has a function of converting an input audio signal into a digital signal and recording the digital signal. Each of the boards 21 to 23 is connected to an internal bus 24, and an audio signal recorded by the A / D conversion board 23 is transmitted to the CPU board 21 via the internal bus 24.

The CPU board 21 executes a process of recognizing the transmitted audio signal, retrieves response information stored in the HDD 25 based on the recognition process, or executes the above-described recognition process on a network. And transmits a response command (recognition result) from the server. Then, the CPU board 21
Sends the response information from the HDD device 25, the response command from the server, and the like to the voice synthesizer 26, where the response information and the response command are converted into a synthesized voice signal. The converted voice signal is transmitted to the telephone line via the telephone answering board 22.

Next, details of the telephone answering board 22 will be described with reference to FIG. In FIG. 2, 220 is LS
I is a digital time-division space switch for switching the internal channel constituted by I.
P221, a network control unit (NCU) 222 described in detail later, an auxiliary input terminal 223 to which an audio signal from the audio synthesizer 26 is input, and an auxiliary output terminal 224 for supplying the audio signal to the A / D conversion board 23 are connected. You. 225 and 226 are telephone line connection terminals. These connection terminals 225 and 226 are connected to the NCU 222.

Reference numeral 227 denotes a connection terminal of the attached telephone. The DSP 221 is connected to the internal bus 2 via the DPRAM 228.
4 is connected. The CPU 211 and the DPRAM 21
2. The DRAM 213 is mounted on the CPU board 21. Reference numeral 214 denotes a connection bus with other devices. Also, NC
U222 is referred to as a network terminating device, and is a lowermost control device in the telephone network. The U222 exchanges information with an exchange installed in a telephone office or the like according to a predetermined procedure and organically connects the terminal device to the telephone network. is there.

As described above, the voice recognition / response device is a CP
Since the three boards of the U board 21, the telephone answering board 22 and the A / D conversion board 23 and the HDD device 25 are housed in a housing, and the voice synthesizing device 26 is provided in the housing,
The overall device has a relatively compact configuration. Further, since each device operates independently, even if a line used for recognition is added, complicated work for the addition is not required, and since the device is compact, the scale of the entire system becomes too large. Expansion can be done without any problem.

Next, a second embodiment of the present invention in which a client-server system is configured using the above-described first embodiment will be described with reference to FIG. In FIG. 3, portions surrounded by broken lines are clients A and B.
B is the voice recognition / response device shown in FIG. The clients A and B are connected to a network 31, to which a server 32 having a manager function for monitoring the clients A and B is connected. The server 32 starts a manager program (the function of this program will be described later) and monitors the clients A and B. The clients A and B are provided with a timer function for notifying the server 32 of the current situation at regular intervals, and are executed in multi-threads in parallel with the recognition process.

The timer function implements a state notification function as described later. This status notification function is
2 in the manager program. Call it on the client side and give it an argument. As a result, the manager program can sequentially check the current state of the client and the speech recognition result. The server 32 not only obtains the client status and the like using the status notification function shown in FIG.
A designated return value can be given to the timer function of each of the clients A and B. The clients A and B refer to the return value from the timer function, recognize the instruction from the server 32, and perform necessary processing. With this function, for example, the client B specified by the server 32 can be shut down or the recognition / response program of the client B can be terminated, and the server 32 can control the entire system.

The processing of the recognition / response program in the second embodiment configured as described above will be described with reference to FIG. The processing of the timer unit, the processing of the push tone / line disconnection detecting unit, the processing of the recognition processing unit, and the like, which are the respective processings in FIG. 5, will be described with reference to FIGS.

First, in the flowchart of FIG. 5 showing the flow of processing of the recognition / response program, the application is initialized in step S1, the manager program of the server 32 is accessed, and the manager program is connected in step S2. Thereafter, in step S3, the voice synthesizer 26 is initialized, and the telephone incoming detecting unit is set in step S3.
Start with 4. If you start incoming call detection,
After the initialization of the telephone line is performed in step S5, the timer unit is started in step S6. When the timer section starts, it waits for an incoming call (step S7), and determines whether there is an incoming call in step S8. As a result of the judgment, "NO"
If so, the process returns to step S7, and if "YES", in step S9, the process of the push tone / line disconnection detecting unit is started, and the process of the recording / recognition processing unit is executed in parallel.

FIG. 6 is a flowchart showing the flow of processing of the timer unit. In FIG. 6, if the timer is started in step S11, the state of the client is notified to the manager program of the server 32 at regular intervals in step S12. . Thereafter, in step S13, a corresponding process is performed with reference to the return value of the status notification function from the server 32. This process is repeated until the timer process ends in step S14.

FIG. 7 is a flow chart showing the flow of processing of the push tone / line disconnection detecting unit. In FIG. 7, line disconnection detection is started in step S21, and line disconnection check is performed in step S22. If the line is not connected in this check, the processing of the recognition processing section is terminated in step S23, the processing of the timer section is terminated in step S24, and the incoming call is awaited again (step S25). On the other hand, if the line is connected in the above check, the detection of the push tone is started in step S26, and if the detection of the push tone is not detected for a predetermined time in step S27 (NO), the process proceeds to step S21.
Return to If a push tone is detected in step S27 (YES), processing corresponding to the tone received in step S28 is executed.

FIG. 8 is a flowchart showing the flow of the processing of the recognition processing unit. In FIG. 8, after the initialization of the recognition processing is performed in step S31, voice input (recording) is started in step S32, and step S33 is started. Perform recording and recognition processing. If the voice input is completed in step S34 (YE
S), the recording and recognition processing ends in step S35, and the recognition result is output to the speech synthesizer 26 and the like in step S36.

In the processing of the recognition / response program, voice is recorded in a recording buffer group. Each time a voice is recorded in one buffer, the buffer is transmitted to the recognition processing unit, and a sequential processing method is adopted.
Since the recording process and the recognition process are performed in parallel, there is almost no waiting time from the end of the voice utterance to the output of the recognition result, and a high-speed recognition process is realized.

The manager program in the server 32 sequentially checks and records the current state, the recognition result, and the like for all clients on which the recognition / response program has been started. This program provides a status notification function to the recognition / response program of each activated client. The status notification function not only obtains the status of the client at each point in time (such as during line connection or voice recognition) and the voice recognition result, but also gives a return value, and provides the client A, B from the server 32. You can tell the order to. This return value is
The setting can be made by selecting a menu prepared on the window of the manager program.

Here, the flow of processing of the manager program will be described with reference to the flowchart shown in FIG. In FIG. 9, the recognition / recognition of an arbitrary client is performed in step S41.
When a connection request is received from the response program, the program is started, and if "YES", a recording unit is created corresponding to the client (S42). When the number of connected clients increases, recording units corresponding to the clients are sequentially created. If the recording unit has been created, it is determined in step S43 whether the command is a client processing instruction by the user. If "YES", a return value to the state notification function is set in step S44. afterwards,
In step S45, the status of each client is received via the status notification function at regular time intervals, and when a menu is selected, a return value corresponding to the menu is transmitted to the status notification function called from the corresponding client and transmitted to the recording unit. Check (S46). After this check, it is confirmed that all connected recognition / response programs have been completed (47), and the manager program is terminated.

[0027]

As described above, according to the present invention,
Because the voice recognition / response device is compact, even if the number of lines to be recognized increases, it is possible to prevent the system from becoming too large.Also, a new type of voice / response device is used to configure the client / server type. There are various excellent effects such that the processing results of the recognition client can be handled collectively and the control becomes easy.

[Brief description of the drawings]

FIG. 1 is a configuration explanatory diagram of a voice recognition / response device according to a first embodiment of the present invention.

FIG. 2 is a diagram illustrating a detailed configuration of a telephone answering board.

FIG. 3 is a configuration diagram of a client-server system according to a second embodiment of the present invention.

FIG. 4 is an explanatory diagram of how to use a state notification function.

FIG. 5 is a flowchart showing processing of a recognition / response program.

FIG. 6 is a flowchart illustrating processing of a timer unit.

FIG. 7 is a flowchart illustrating processing of a push tone / line disconnection detection unit.

FIG. 8 is a flowchart showing processing of a recording / recognition processing unit.

FIG. 9 is a flowchart showing processing of a manager program.

FIG. 10 is a schematic configuration diagram of a conventional voice recognition device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 21 ... CPU board 22 ... Telephone answering board 23 ... A / D conversion board 24 ... Internal bus 25 ... HDD device 26 ... Voice synthesizer 31 ... Network 32 ... Server

Continued on the front page F term (reference) 5D015 KK01 5D045 AB24 AB30 5K015 AA00 AA06 AA07 GA00 GA07 5K024 AA76 BB01 BB02 CC01 DD01 EE01 EE09 FF06

Claims (3)

[Claims] 1. A telephone answering board equipped with a CPU board on which various memories are mounted, an A / D conversion board for converting an analog signal into a digital signal, a network terminator and a digital switch, and provided with a telephone line connection terminal. Board and
A conductor for electrically connecting each of these boards;
A voice synthesizer for generating a signal output from the U board into a synthesized voice signal and supplying the synthesized voice signal to the telephone answering board, and an analog signal transmitted from a telephone line connected to a telephone line connection terminal of the telephone answering board. A signal is transmitted from the telephone answering board to the A / D conversion board, converted into a digital signal by the A / D conversion board, and the digital signal is converted
A telephone voice recognition / response device, wherein the voice data is transmitted to a board, a recognition process is executed, the recognition result is converted into a synthesized voice by a voice synthesizer, and then transmitted to a telephone line from a telephone response board. 2. A telephone voice recognition / response apparatus according to claim 1 is connected to a plurality of networks as a client, and a recognition result is transmitted to a server via the network, or a command from the server is received. A telephone voice recognition system wherein a client automatically performs a corresponding process. 3. A recording medium storing a program for controlling a telephone speech recognition system, comprising: means for using a status notification function provided by a server; A recording medium storing a telephone speech recognition system control program, comprising: a program for notifying the above command via a return value; and a program functioning as a program for controlling the client from the server.