JPS60106260A - Speech connecting device of automatic exchange - Google Patents

Abstract

PURPOSE:To operate easily various indoor exchange functions by using voice by providing an automatic exchange with a specific speaker speech recognizing device and a speech synthesizer. CONSTITUTION:When a tone signal generated with the specific key of a subscriber telephone set 1 or 2 is inputted, the specific speaker speech recognizing device 6 recognizes a voice inputted within a specific time through a timer gate part 5. The device 6 analyzes words of an unspecified speaker into a sequence of spectrum information to peform matching 8 corresponding to a word dictionary 9 according to respective functions, and narrows down results to several candidates, deciding that the 1st candidate is more matched than any other candidate. When the degree of matching is lower than a threshold value, the speech synthesizer 12 generates a synthesized voice to the subscriber and makes a connection to a transfer destination according to a transfer destination telephone number sent from the telephone set 1 or 2.

Description

[Detailed Description of the Invention] Industrial Field of Application The present invention is characterized in that a residential automatic switchboard is equipped with a speaker-independent word recognition device and a speech synthesis device, and a line is fully automatically connected using the subscriber's voice from the subscriber telephone. This invention relates to a voice connection device for an automatic exchange that allows users to easily perform various line connection functions by instructing the setting of each function button of a telephone using a voice synthesizer.

Conventional configurations and their problems In recent years, in-home exchanges for telephones have more complex functions, and efforts have been made to improve the efficiency of offices and the like. In addition, these functions include calls such as call forwarding, which automatically forwards calls to the new location when the user is absent and goes to another department, and calls that can be made simultaneously by multiple subscriber telephones, allowing multiple people to make calls at the same time. It has functions such as conference calls.

However, when using these functions, complicated operations such as numeric keys 1 to O and special keys such as "US" and "#" are required, resulting in poor usability. Furthermore, voice recognition technology, such as that described in Japanese Patent Application Laid-Open No. 56-64587, is installed in telephone exchanges to record telephone numbers and their corresponding names, addresses, etc. Devices that perform number guidance do not have the above-mentioned various functions of telephones, and they also have the disadvantage of causing malfunctions due to erroneous recognition.

Purpose of the Invention The present invention has been made in view of the above-mentioned drawbacks, and is equipped with a voice recognition device and a voice synthesis device in an automatic switchboard, etc., and solves the operational difficulties that were a problem with conventional automatic connection of telephones by key operation. It is an object of the present invention to provide a voice connection device for an automatic exchange, which can eliminate noise, improve operability, and prevent erroneous connections due to erroneous voice recognition.

Structure of the Invention In order to achieve the above-mentioned object, the voice connection device of the automatic exchange of the present invention allows each telephone subscriber to perform automatic transfer in the event of an absentee call, hold the line on hold, call all subscriber telephones simultaneously, and use multiple subscriber telephones that are set. Automatic telephone and data exchanges with functions such as voice conferences are equipped with a speaker-independent word recognition device and a speech synthesis device, and the word names for each function are input using a voice input device such as a microphone installed in a handset or telephone. is input as a voice, the speaker-independent word recognition device automatically connects the line corresponding to the function, and the voice synthesizer sends synthesized sounds corresponding to each connection function to the subscriber telephone. This makes it possible to operate various home exchange functions easily and without erroneous connections using voice.

DESCRIPTION OF THE EMBODIMENTS FIG. 1 is a block diagram of a voice connection device for an automatic exchange according to an embodiment of the present invention.

In FIG. 1, 1 and 2 are subscriber telephones installed at different locations, 3 is a network, 4 is a network control device, and 5 is a voice recognition setting signal input from subscriber telephone 1 or the user's transferred word voice. 6 is a speaker-independent speech recognition device, and an acoustic analysis section 7. Matching determination section 8. It is composed of a word dictionary section 9 having numerical sound patterns. 1o is the judgment result of the unspecified speaker speech recognition device 60 and the timer gate unit 5
13 is a part of the synthesized sound data memory;
14 is a composition control section.

The present embodiment constructed as described above will be described below with reference to FIGS. 1 to 3. FIGS. 2 and 3 are flowcharts of the processing operations of this embodiment.

An example of missed transfer will be described in which all calls made to the subscriber telephone 1 are transferred to the subscriber telephone 2 in the conference room when the user of the subscriber telephone 1 is absent and goes to a conference room or the like.

After the subscriber's line is connected to the automatic exchange, a tone signal from a telephone or the like or a separately provided key is detected to determine whether or not voice recognition is to be performed. (Step 16) Here, if the tone other than the key operation is a tone other than the key operation, the normal dial operation is performed, and the network control device 4 connects the predetermined network 3 (Step 16). If a signal or the like is input at tL7, a timer performs a voice recognition operation within a certain period of time (step 17).

The monophonic voice recognition section targets word recognition for unspecified speakers, performs acoustic analysis on a series of spectrum information using filter groups, Fourier transform, LPG, etc., and performs matching complete matching corresponding to academic speech according to each function. As a result, a plurality of candidates are selected, and it is determined that the first candidate matches better than the other candidates.

(Step 18) As a result of the recognition, it is determined whether the degree of matching falls within a certain threshold (Step 19), and if it is below the threshold IW, a synthesized monotone is generated for the subscriber (Step 20), the network control device 4, a connection is made to the forwarding destination according to the forwarding destination telephone number sent from the subscriber telephone. (Step 21
). If the threshold value is exceeded, a synthesized voice requesting the subscriber to speak again is generated (step 22), and the voice re-input within the timer set time (step 23) is recognized. (Step 24). Further, for the N times of recognition operations set by the counter, the subscriber is requested to re-enter all the information until the loop reaches N times.

(Steps 2a, 26). Here, if matching is obtained within the threshold (step 27), as in step 19 described above, a synthesized voice informing the subscriber of absent transfer is sent to the subscriber as the transfer destination, and all connection operations are performed (step 28. 29)
.

Also, if it is impossible to determine even if the counter reaches a predetermined number of times N, among a large number of multiple candidates.

The hogo with the highest score is weighted in order from the first candidate. It is set as a recognition word, and the score is compared with the next candidate word, and the score is lower than the set value (discrimination is impossible).
It is determined whether or not (step 30.31). As a result, if there is a significant difference between the first and second candidates, a synthesis event will occur (Stesoge28), and if it is impossible to distinguish, and the subscriber's voice comes in from the first to several candidates. "Yes","
Recognize one of the two words ``Yes'' (steps 35 and 36), and perform recognition operations until the final candidate word is ``Yes''.

Performs two-word recognition operation of “Yes” (step 37)
, when all voices of "hi" are recognized, a synthesized voice corresponding to a predetermined function (for example, "call forwarding") is generated and a line connection operation is performed (steps 38 and 39).

If the final candidate words are not recognized, or if the candidates cannot be ranked, the subscriber will be sent an operating procedure to the IB voice synthesizer 12, and the subscriber will be able to connect the same way as a conventional line connection. (Step 40).

As described above, according to this embodiment, by providing the automatic switchboard with the speaker-independent speech recognition device 6 and the speech synthesis device 12, various home switching functions can be performed without complicated operations such as special keys. It can be easily operated using voice, and the word dictionary has numeric and voice patterns, so you can use only voice to give instructions on forwarding destinations, as well as the attributes of each subscriber's phone number, such as their name and affiliation. Not only does this enable automatic forwarding settings based on the name and affiliation of the person being forwarded to, making it easier to operate, but it also allows for a variety of functions such as putting the line on hold and connecting to Parlor Line for conference calls. Since a functional line connection can be made while confirming voice recognition, incorrect line connections can be completely prevented.

Effects of the Invention As described above, according to the voice connection device for an automatic exchange of the present invention, by providing the automatic exchange with a speaker-independent voice recognition device and a voice synthesis device, complicated operations such as special keys can be performed. It can be easily operated using various home exchange functions (kW voice), and the word dictionary has numeric voice patterns, so you can use only voice to instruct forwarding destinations, name and name of each subscriber, etc. By adding a phone number attribute to affiliation, etc., it is possible to automatically set up forwarding based on the name and affiliation of the person being forwarded to, which not only makes operations easier, but also allows for easier operations such as putting the line on hold or holding a conference call. Since multifunctional line connections such as Parlor line connections can be made while confirming voice recognition, erroneous line connections can be prevented, and this has a great effect.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a voice connection device of an automatic exchange according to an embodiment of the present invention, and FIGS. 2 and 3 are flowcharts of the same processing operation. 1.2... Subscriber telephone set, 3... Network, 4... Network control device, 6...
. . . Speaker-independent speech recognition device, 12 . . . Speech synthesis device. Figure 1 Figure 2

Claims (1)

[Scope of Claims] (1) It has functions such as automatic transfer from the valley subscriber telephone when absent, holding the line on hold, simultaneous ringing of all subscriber telephones, and voice conference using the set telephone number 7In. Automatic telephone and data exchanges are equipped with an unspecified speaker word recognition device and a cry synthesis device, and the word names of each function are input aloud using a voice input device such as a microphone installed in a handset or a telephone set, and unspecified talk can be performed. A voice connection device for an automatic exchange, in which a voice recognition device automatically connects a line corresponding to the function, and a voice synthesis device sends a synthesized voice corresponding to each connection function to subscriber telephones. (2) The automatic exchange operates the speaker-independent voice recognition device in accordance with the recognition setting switch signal sent from each subscriber telephone, and then uses the Rin language corresponding to each connection function of the input user to input the next input. After the voice recognition device recognizes the voice and before connecting the line, a synthesized voice corresponding to the recognized word is sent to each subscriber telephone set, and based on that, a confirmation voice is re-inputted by the subscriber's voice to create a hierarchical structure. A voice connection device for an automatic exchange according to claim 1, which connects a subsequent line by performing a recognition operation. (3) If the speech recognition device makes a false recognition, sends a synthesized speech corresponding to the second candidate word obtained in the speech recognition device to the subscriber telephone, and requests confirmation speech from the subscriber;
Claim 2 of connecting the line by the recognition operation again
The voice connection device for the automatic exchange described in Section 1. (4) If the subscriber voice input signal input to the voice recognition device cannot be recognized and discriminated from the primary candidate to the multiple candidates as a result of voice recognition, a request is made to re-input the subscriber voice from the voice synthesizer. 2. The voice connection device for an automatic exchange according to claim 1, wherein the voice connection device for an automatic switching exchange transmits a message to confirm the connection function requested by the subscriber. (6) If the voice recognition device is unable to recognize and distinguish even if the voice of the connection function is input using the subscriber voice multiple times, among the multiple candidate words obtained from the voice recognition device based on the subscriber voice multiple times. 5. The voice connection device for an automatic converter according to claim 4, wherein synthesized voices are transmitted by a voice synthesis device in order of the number of recognitions performed by the subscriber, and the line is connected after the subscriber's voice is confirmed by the voice recognition device. (6) Voice: If the recognition device is fully capable of recognizing and discriminating, the voice synthesizer will send synthesized voices to the subscriber telephone to explain the operating procedures corresponding to each line connection, and the subscriber will be able to respond to each function. A voice connection device for an automatic exchange according to claim 1, wherein a button is set and a line is connected based on the signal thereof.