JPH0618395B2 - Voice dial device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a voice dial device in a telephone, and more specifically, a voice to be recognized is registered in advance, and the voice is input through a voice input means. The present invention relates to a voice dial device including voice recognition means for recognizing an input voice by comparing it with a registered voice and dial signal transmitting means for transmitting a dial signal through a telephone line based on the recognition result.

[Conventional technology]

A call operation (dial) operation in a normal telephone is performed by a person lifting a handset, connecting a telephone circuit to a telephone line, and operating a rotary dial or a push button with a finger to make a dial signal consisting of a string of about 10 digits. Was sent to the line.

As described above, in the calling operation in the ordinary telephone, the complicated operation of manipulating the buttons and rotary dials corresponding to the numbers by hand is required about 10 times. Therefore, when the hand is inconvenient, for example, when the user wants to make a call while driving a car like a car telephone, the hand is not free and inconvenient.

Japanese Patent Laid-Open No. 59-22565 discloses a technique for solving such a drawback and for making a telephone call without using a hand when inconvenient.
There is a so-called voice dial device using a voice recognition technique as described in Japanese Patent Laid-Open No. 6 or Japanese Patent Laid-Open No. 60-59846. This is to automatically send a dial signal by voice and does not require manual dial operation.

[Problems to be solved by the invention]

In the above-mentioned conventional technology, since automatic dialing is performed by voice, the destination to be dialed is first input to the voice recognition device by voice. However, the voice recognition device accepts the voice input and its period, that is, what time. It starts accepting voice input from the point of time, and no consideration is given to when it ends, and for example, it is possible to always accept voice input as a voice recognition device,
If you are ready to receive voice input at any time and are constantly in a standby state, especially in the case of a mobile phone, the ambient noise is erroneously input as voice input and the voice recognition device malfunctions. However, there is a problem that an unintended dialing operation may occur.

It is an object of the present invention to reasonably automatically set a voice input acceptance timing for dialing and the length of a period in dialing by voice recognition, and perform voice recognition by cutting out a voice input within this period. By doing so, the probability that inadvertent ambient noise is input to the voice recognition device is minimized to prevent malfunction of the voice dial device.

[Means for solving problems]

The purpose is to detect that the voice input means to the voice recognition device, for example, the handset transmitter approaches the caller's mouth (in this case, the receiver inevitably approaches the ear, which may be the case). Proximity detection or means for detecting that a certain period of time has elapsed from when the handset is lifted until it approaches the mouth or ear of the caller, and a standard registered in advance in the voice recognition device for voice recognition. A voice length measuring means for measuring a voice length of a voice (registered voice) in advance, and an overall statistical property of the measured voice length, for example, a maximum value, an average value, and a variance value are statistically processed and derived, A voice length statistical means for determining a reasonable time length (specified time) required for recognition and a control means for controlling the voice recognition device are provided, and the voice reception time for dialing is controlled by the control means. Inspection Means mouth transmitter (ear receiver)
This is achieved by limiting the time from the time when it is detected to approach to the predetermined time determined by the statistical means, for example, the time obtained by adding the voice length variance value to the voice length average value by 3 times.

[Action]

The proximity detection means detects that the caller has brought the transmitter (microphone) to the vicinity of the mouth, because the caller inputs the voice uttered by the caller to dial the voice recognition device. The voice length measuring means measures the time length of the entire registered voice that the caller previously inputs to the voice recognition device for voice recognition and registers as a standard voice. The voice length statistical means performs statistical processing on the entire measured voice length, derives its statistical properties such as maximum value, median value, average value, variance value, etc. Time).

The control means limits the inputtable time to the voice recognition device based on the detection time of the detection output by the proximity detection means and the specified time determined by the statistical means until the specified time elapses from the detection time. Control the operation.

As a result, the voice recognition device can limit the reception time of the input voice for dialing, reduce the probability of accidental input of inadvertent ambient noise, etc., and prevent malfunction of the voice dial device.

〔Example〕

An embodiment of the present invention will be described below with reference to FIG. First
FIG. 1 is a block diagram showing a telephone circuit having a voice dial device as an embodiment of the present invention.

In FIG. 1, (1-1) and (1-2) are line terminals,
2 is a handset, 3 is a handset built in the handset 2, 4 is also a transmitter, 5 is an electrically controlled automatic hook switch for opening and closing the telephone circuit, and 6 is a two-wire 4 An anti-side sound circuit that performs line conversion, 7 is installed near the transmitter 4 in the handset, and a proximity switch that detects that the mouth of the operator (caller) is close to the transmitter 4 is indicated by 8 A dial signal generating (sending) circuit for sending a dial signal to the line, and 9 detects a calling signal called from an exchange (not shown) by inputting through the line terminals (1-1) and (1-2). Incoming call detection circuit, 10 is a bell sound generation circuit that generates a bell sound to notify an incoming call by an audible sound according to a call signal, 11 is a bell sound speaker, 12 is a control circuit for controlling the whole, 13 is a voice recognition circuit, 14 is the time of each registered voice Is a voice length measurement circuit for measuring the voice registration time, 15 is a voice length memory that stores each time length of the registered voice that is measured, 16 is a maximum value of these based on the measured voice length data, A voice length statistical circuit that calculates a value indicating statistical properties such as a median value, an average value, and a variance value and defines a reasonable time length for voice recognition based on the calculated value.
Is a specified time register for storing and storing a specified time specified by the voice length statistical circuit 16, 18 is a dial push button for inputting a dial telephone number, 19 is an input voice (for example, the name of the other party) and a dial number transmitted. Is a registration switch used for pre-registering the above relationship, 20 is a hook detection circuit for detecting the lifted state (off-hook) of the handset 2, and 21 is a mixing circuit.

Next, the operation of the voice recognition circuit 13 will be described.

This voice recognition circuit is a circuit that is currently in practical use, but is not limited to this circuit for recognizing a word voice that a specific speaker divides and utters. First, in order to enable voice recognition, it is necessary to register the voice of the operator (specific speaker) in advance. This is called voice registration. This voice registration is an operation in which the voice recognition circuit 13 is set to a voice registration operation mode and the input voice is sequentially stored as a standard pattern in a standard pattern memory (not shown) in the circuit 13.

For example, when the utterance "Suzuki" is made from the transmitter 4 to the voice recognition circuit 13, this utterance is cut out by a voice cutout circuit (not shown) in the circuit, and this feature is analyzed and extracted to make a time-series feature. It is stored in the standard pattern memory as a parameter pattern. "Sato", "Ota" in sequence ...
Each time a voice is uttered, each standard pattern is additionally stored in the standard pattern memory in the circuit 13.

Next, the voice recognition operation will be described. The voice recognition circuit 13 is set to the recognition operation mode, and the same voice as previously registered, for example, "Suzuki" is input. This voice is cut out as in the case of voice registration and is temporarily stored in an input pattern memory (not shown) in the circuit 13 as a characteristic parameter pattern. Next, the contents of this input pattern memory and the contents of each standard pattern in the standard pattern memory are sequentially compared, and the information about which standard pattern was most similar is output as a result. In this case, the standard Suzuki code pattern code is output. As the comparison method, a known absolute value distance calculation method and a similarity calculation method such as a DP matching method are used.

From the above, it is possible to understand the voice registration operation necessary for voice recognition and the recognition operation of the voice input thereafter.

The overall operation of the telephone circuit shown in FIG. 1 will now be described with reference to the flow chart of FIG.

Starting from turning on the power, as shown in step S1, it is determined whether or not the registration switch 19 is pressed,
If it is pressed, the operation proceeds to the registration operation (step S2). If it is not registered, the process moves to step S3 to detect a call (incoming call) by the incoming call detection circuit 9, and if there is an incoming call, a bell sound is produced by the bell sound producing circuit 10 and the bell sound speaker 11. At the same time as notifying the operator, the control circuit 12 also knows this.

When there is an incoming call, the operator hears the bell sound and lifts the handset 2 (step S4). Then, the hook detection circuit 20 detects that the handset 2 is off-hook, and the control circuit 12 controls the automatic hook switch 5 to close it, and the telephone circuit is set to the line terminals (1-1), (1-).
Connect to the line via 2). Then, the operator enters a call operation (step S5). When the call ends and the operator puts the handset on, the hook detection circuit 20 detects this, and the control circuit 12 controls the automatic hook switch 5 to open and disconnect the telephone circuit from the line.

When there is no incoming call and the handset 2 is lifted in step S6, the operator dials the other party, that is, enters into a calling operation, and the telephone circuit shifts to a calling operation (step S7). This calling operation will be described later.
When the calling operation is completed, the call operation is continued (step S5)
Then, the operator talks with the dialed party.

Next, the registration operation of the voice for automatic dialing by voice recognition and the corresponding telephone number will be described. FIG. 3 is a flow chart of this registration operation.

Please refer to FIG. 1 and FIG. The operator switches the registration switch 19
Press to enter the registration operation. Then, when the operator lifts the handset 2 and brings the transmitter 4 in the handset near the mouth, the proximity switch 7 detects this (step S1). Then, the control circuit 12 controls the voice recognition circuit 13 so that a voice synthesis circuit (not shown) in the circuit 13 utters, for example, “ONSEIYONYOKURIKUKUTEKUDAKU” and outputs it from the handset 2 via the mixer 21. Then, the operator is notified of the input (step S2). After that, the voice recognition circuit 13 is set to the voice registration operation mode.

According to this notification, the operator utters the name of the other party who automatically dials, for example, "Suzuki" (step S3). This voice is stored in the standard pattern memory in the voice recognition circuit 13 as "00".
Is stored as a standard pattern. Then, the voice length of this standard pattern is measured by the voice length measuring circuit 14 (step S5), and this value is stored in the voice length memory 15
(Step S6). This voice length is the time from the start of the cutout by the cutout circuit to the end thereof.

The operator inputs Mr. Suzuki's telephone number, for example, "100", using the dial push button 18 (step S
7). This telephone number is stored in a telephone number memory (not shown) in the control circuit 12. At this time, the code "00" of the standard pattern is also stored in the telephone number memory as a pair (step S8). Next, the control circuit 12 causes the voice length statistical circuit 16 to obtain a value indicating the statistical property of the voice length data of the standard pattern in the voice length memory 15, for example, a maximum value, a median value, an average value, and a variance value. Etc. are calculated (step S9). In this case there is still only one voice length,
This voice length is the maximum value, the median value, and the average value. Then, the specified time is calculated based on this value. For example, a value that is twice the maximum value is set as the specified time and written in the specified time register 17 (step S10). The calculation of the specified time is not limited to twice the maximum value as described above.

When further registering, the registration switch 19 may be left as it is and the above sequence may be continued (step S11,
S12). For example, the name of the person to auto dial
For example, say "Sato." The voice "Sato" is stored as a standard pattern with the code "01" added to the standard pattern memory. The voice length is also stored in the voice length memory 15. Then, the operator inputs Mr. Sato's telephone number, for example, "101" using the dial push button 18. This telephone number is likewise stored in the telephone number memory together with the standard pattern code "01". And the voice length of "Suzuki" and the current "Sato"
The voice length statistical circuit 16 calculates a value indicating the statistical property by using the voice length of, the prescribed time is calculated, and the prescribed time is written in the prescribed time register 17. That is, the specified time is updated at this point.

Hereinafter, the operator registers as many names as necessary. Based on the voice length data of the number of voices registered so far, a value indicating the statistical property is calculated, and then the specified time is constantly updated and obtained.

When the registration is completed, the operator pushes the registration switch 19 to leave the registration operation.

Here, the advantage of the proximity switch 7 is described. When the proximity switch 7 is turned on and the mouth is brought near the transmitter 4,
The operator does not hear the notification sound from the receiver 3 because the input notification sound is emitted. Also, until the handset 2 is brought close to the mouth, the voice recognition circuit 13 does not enter the operation for voice registration, and an unintentional ambient sound, for example, a mechanical sound produced when the handset 2 is lifted is input. Nothing.

In the operation flow of FIG. 3, for each registration, a value indicating the statistical property is calculated based on the voice lengths registered so far, and the specified time is updated. After the registration, for example, when the registration of 10 persons is completed, the values may be collectively calculated based on the voice lengths of the 10 persons, and the specified time may be updated. That is,
The statistical value calculation (step S9) and the specified time writing (step S10) in FIG. 3 may be performed after the series of registrations.

FIG. 4 shows the operation flow in this case. It is not necessary to explain FIG.

Also, in the above, only a series of registration operations were explained, but if the registration operation is ended and a new person's name and telephone number are newly registered several days later, the same operation may be performed. The voice length data of the last registered voice is retained,
It is apparent that the voice length for additional registration is further added to this data, and the specified time is updated based on the voice length data accumulated so far.

In addition, it may be possible to delete the existing registration, and this deletion can be performed by deleting all the voice pattern data corresponding to the standard pattern and telephone number of the designated name. And
At this time, it is necessary to recalculate the value indicating the statistical property in the voice length statistical circuit 16 again by using the data that has been erased and excluding the voice length data, obtain the specified time, and update the specified time register 17.

Further, the input notification may be an electronic sound, for example, a beep sound, instead of the voice of the voice synthesis.

Next, a calling operation, that is, an automatic dialing operation by voice recognition will be described. FIG. 5 is a flow chart of this calling operation.

When the operator lifts the handset 2 to dial, the hook detection circuit 20 detects that the handset 2 is off-hook. At this time, the incoming call detection circuit 9 has not detected an incoming call.
Under these two conditions, the control circuit 12 recognizes that the operator is trying to make a call, and shifts the telephone circuit to the calling operation mode.

First, the operator holds the handset 2 near his mouth.
Then, the proximity switch 7 detects this (step S
1), the voice recognition circuit 13 is set to the recognition operation mode and stands by in preparation for voice input from the transmitter 4. The operator utters the name of the other party who registered the voice, for example, "Suzuki" before attempting to make a call (step S2). This utterance is the transmitter 4
Is picked up and input to the voice recognition circuit 13.

The voice recognition circuit 13 takes in this voice, and inputs the input voice pattern “Suzuki” and the standard patterns such as “Suzuki”, “Sato”, “Ota”, etc., which are registered in advance, as described in the previous voice recognition operation. As a result of comparison, the code of the most similar standard pattern, in this case the code "00" (indicating the standard pattern of "Suzuki"), is output to the control circuit 12 (step S3). At this time, the control circuit 1
2 is based on the value of the specified time register 17, if the voice input, that is, the voice recognition circuit 13 does not cut out the voice within a specified time after the proximity switch 7 is turned on (step S4), the voice recognition is performed once. The recognition operation of the circuit is stopped, and a voice synthesizing circuit (not shown) in the circuit utters an input notification, for example, "ONSEIYONYORIKUKUKU TEKUDA Sai".
The name is output from the handset 3 via the mixer 21, and the operator is prompted to input a name (step S5).

The same applies when voice input is not completed within the specified time. If it is completed within the specified time, the next operation is performed.

Since there is a misrecognition in voice recognition, it is necessary to confirm the recognition result. Therefore, the control circuit 12 uses the voice recognition circuit 13 to confirm and present the voice "Suzuki" corresponding to the standard pattern code, which is the result output, to the operator using the synthesis circuit. For example, ask for "high" and "yes". After that, the control circuit 12 sets the voice recognition circuit 13 in the recognition operation mode and stands by for the next confirmation utterance.

The operator listens to this voice on the handset 3, and if it is correct, speaks "high", and if it is wrong, speaks "yes" (step S7). Naturally, the utterances of "high" and "yes" also need to be registered in advance. When the operator confirms that it is correct and inputs "high", the code of "00" and the code of "high" are output to the control circuit 12 again as a recognition result.

At this time also, as in the case of the name utterance, the control circuit 12
Is based on the value of the specified time register 17, if there is no voice input within a specified time after the confirmation voice is synthesized, or if the input is not completed, the voice recognition operation is temporarily stopped, and The user is notified of the input and prompted to input the confirmation utterance again (step S10).

In the case of "yes", the code of "yes" is output to the control circuit as the recognition result. Then, the user is again instructed to utter the name again (step S12), and the control circuit 1
2 controls the voice recognition circuit 13. In the recognition of "high" and "yes", if the number of categories, that is, the standard patterns to be recognized is limited to two, "high" and "yes", a very high recognition rate can be obtained, so that the name utterance There is no need to check the result like with voice recognition.

In the case of "high", the control circuit 12 controls the automatic hook switch 5 based on the output code of "high" to close it and connect the telephone circuit to the line. After that, referring to the code "00", the telephone number "100" of Mr. Suzuki stored in the telephone number memory is obtained, the dial signal generation circuit 8 is controlled, and the dial signal "100" is sent to the line.

FIG. 6 is a flowchart of another calling operation of this embodiment. Fifth
In the figure, the handset 2 is lifted, brought near the mouth, and when the proximity switch 7 detects this, the voice recognition circuit 13 is set to the recognition operation mode, and the voice recognition circuit 13 waits for voice input and monitors for a specified time. However, in FIG. 6, after the proximity switch 7 detects the approach of the mouth, an input reminder is issued, and thereafter, the specified time is monitored. By doing so, the user knows when to occur and does not overlook the lapse of the prescribed time unnecessarily. With respect to FIG. 6, there is no need for further explanation.

FIG. 7 is a flow chart of another calling operation in the implementation order of FIG. The flowchart of FIG. 7 differs from the flowchart of the operation shown in FIGS. 5 and 6 in the operation for confirmation. Fig. 5,
In Fig. 6, for confirmation, the voice that corresponds to the most similar standard pattern is synthesized, output for confirmation, and "High" and "No".
In the case of “yes”, that is, when the confirmation voice is recognized, if it is incorrect, the operator may repeat the name generation by repeating the calling operation from the beginning.

FIG. 7 eliminates this inconvenience. In the case of "No", the next (second, usually called second place candidate) similar standard pattern voice is synthesized and the result is confirmed and output again. The recognition of the confirmation occurrence of "high" and "yes" is performed again. This is a method of successively synthesizing the voices corresponding to the next similar standard patterns until the "high" confirmation utterance is made, and repeating confirmation output, confirmation utterance, and confirmation voice recognition. This means that the cumulative recognition rate for voice recognition is always 1
It is based on the principle that the voice corresponding to the standard pattern intended by the operator is always confirmed and output when it becomes 00%, that is, in the order.

According to FIG. 7, the operator can make only one name utterance, and after that, the operator's burden can be reduced by responding with "high" or "yes" to the voices sequentially output by the voice recognition circuit 13. To be done.

At this time, it is not necessary to say "yes". This is considered to be the same as the "yes" utterance if no voice input is made within the input reception time of the voice recognition of the confirmation utterance, that is, the specified time, and the input reminder is not issued. The flow of the broken line in FIG. 7 shows this. In this way, the operator can dial the name of the person to be dialed, listen to the confirmation output in sequence, and dial "high" after the correct confirmation output to dial automatically.

In the flow chart of FIG. 7 as well, as in the flow chart of FIG. 6, the input notification may be given after the determination of the proximity switch ON.

In the above description of FIG. 5 to FIG. 7, the specified time is set regardless of whether or not there is input (whether or not the cutout is started) from the time when the proximity switch 7 detects the approach of the mouth or the time when the first input notification is finished. If monitoring is started, but there is an input within the specified time, monitoring of the specified time is temporarily stopped when this input is cut out, the elapsed time until then is cleared, and monitoring of the specified time is restarted. May be.

As described above, according to the present embodiment, the voice recognition circuit 13 does not enter the input standby state only by lifting the handset 2, and the mechanical sound generated when the handset 2 is lifted is not recognized as a voice and malfunction occurs. To be prevented. Further, since the input time is limited to the time defined by the value indicating the statistical property of the registered voice as described, the probability that an inadvertent ambient noise is input to the voice recognition circuit 13 and malfunctions can be reduced. .

Also, in the explanation, the specified time for utterance of the name and the specified time for utterance of confirmation are the same.
The voice length of "yes" and the voice length of the name utterance are distinguished, the value indicating the statistical property is calculated by the voice length statistical circuit 16 for each group, the specified time is calculated, and the specified time is stored in different specified time registers. If it is stored, the specified time when the name of the calling operation is uttered and the specified time when the confirmation is uttered can be made independent.

In FIG. 1, the dial signal voicing circuit 8 is connected to the line side as the voicing source of the pulse dial signal, but if it is the DTMF signal voicing source, the four side lines of the protection side sound circuit 6 are transmitted. It is preferable to connect the output to the signal path.

In the description, the automatic dialing operation by voice recognition has been described, but naturally the dial push button 18 is manually operated.
It is easy to make dialing by using, for example, a changeover switch for instructing manual / voice dialing.

FIG. 8 is a block diagram showing another embodiment of the present invention.
The same symbols as in FIG. 1 indicate the same things.

In the embodiment of FIG. 8, the proximity switch 7 is removed and a timer circuit 22 is provided instead. Timer circuit 2
2 detects the elapse of a predetermined time, for example, 1 second after the hook detection circuit 20 detects that the handset 2 has been lifted, and notifies the control circuit 12 of it.

FIG. 9 is a diagram showing the flow of the registration operation of the embodiment shown in FIG. The operation flow in FIG. 9 is obtained by replacing the monitoring by the proximity switch in FIG. 3 with the monitoring by the timer circuit 22 for a certain period of time.

When the operator presses the registration switch 19, the registration operation starts. Then, when the operator lifts the handset 2, the hook detection circuit 20 detects this, and then the timer circuit 22 detects that a certain period of time has elapsed, and thereby issues an input. After that, the voice recognition circuit 13 is set to the voice registration operation mode. Since the operation is the same as that shown in FIG. 3 thereafter, the description thereof will be omitted. The fixed time detected by this timer circuit 22 is
This is to allow for the time required for the operator to lift the handset 2 and bring the handset 4 closer to his / her mouth, and this extension of time prevents the operator from hearing the reminder voice from the handset 3. Absent.

Also, the voice recognition circuit 13 does not start the operation for voice registration until the handset is brought close to the mouth, and careless ambient sounds, such as mechanical sounds generated when the hunt set is lifted, are transmitted to the voice recognition circuit. It will not be entered.

It is also apparent that the operation flow shown in FIG. 9 can be modified into the operation flow shown in FIG.

FIG. 10 is a diagram showing the flow of the calling operation of the embodiment shown in FIG. The monitoring of the proximity switch in the operation flow of FIG. 5 is replaced with the monitoring of the elapse of a predetermined time by the timer circuit 22.

When the operator lifts the handset 2 to dial, the hook detection circuit 20 detects that the handset 2 is off-hook. At this time, the incoming call detection circuit 9 has not detected an incoming call.
Under these two conditions, the control circuit 12 recognizes that the operator is trying to make a call, and shifts the telephone circuit to a calling operation. The timer circuit 22 counts the lapse of a fixed time after the detection of the off-hook and notifies the control circuit 12 after the fixed time.

The control circuit 12 puts the voice recognition circuit 13 into the voice recognition operation mode after a certain period of time, and waits to capture the name utterance of the operator. The subsequent operation is similar to that of FIG.

After this fixed time, the voice recognition circuit 13 waits for voice input, so that careless sound generated when lifting the handset is not input to the voice recognition circuit. Also, since the input waiting time is limited to the previously specified time, there is a low probability that careless noise is input to the voice recognition circuit and malfunctions.

It is also apparent that the operation flow of FIG. 10 can be modified as shown in FIGS. 6 and 7.

FIG. 11 is a block diagram showing still another embodiment of the present invention. This embodiment is an embodiment applied to a car telephone.

11, the same symbols as in FIG. 1 indicate the same things. Reference numeral 23 is a wireless antenna for exchanging signals with a base station (not shown), and 24 is a mobile device (for example, an automobile), which is a control signal and voice for performing wireless telephone communication with a base station (not shown). A mobile device that sends and receives signals.
In a car telephone, an incoming signal (calling signal), a dial signal, and a connection with a line by a hook switch in an ordinary terrestrial telephone line are all performed by wireless transmission and reception of the above control signal (called a control signal channel). In addition, the voice signal is assigned to the radio channel for which the reception and the transmission are independent.

The basic operation is similar to that of FIG.
Mobile unit 24, incoming call detection circuit 9, dial signal generation circuit 8
The relationship will be explained. The call signal is coded from a base station (not shown) on a control signal channel and is encoded by the antenna 2
3 is received by the mobile device 24. This signal is detected by the incoming call detection circuit 9, converted into an audible sound by the bell sound generation circuit 10 and the bell sound speaker 11, and is output to the control circuit 12 simultaneously with notifying the operator. Dial signal generation circuit 8
The dial signal of is also encoded, and is transmitted from the movement amount 24 to the base station via the antenna 23 on the control signal channel. In the case of an incoming call as an example of connection to the line, first, a bell sound is generated by an incoming call signal from the control channel, and when the operator lifts the handset 2, the hook detection circuit 20 detects this lifting. However, the control circuit 12 controls the mobile unit 2
4, a code for instructing connection to the line is transmitted to the base station through the control channel. Then, a voice signal receiving / transmitting wireless channel is set between the base station and the mobile device 24, and the call becomes possible.

In this way, in car telephones, the control of the network such as ringing signals, dial signals, and connection to lines is performed by transmitting and receiving wireless signals through the control channel, and voice signals are separate from these and are independent wireless channels for transmission and reception. Except that the side soundproof circuit 6 is not necessary, the same as a normal telephone. Therefore, the registration operation and call origination operation in this embodiment are the same as those in the embodiment of FIG.

The embodiment of FIG. 11 can also be modified like the embodiment of FIG. 8 by replacing the timer circuit with a proximity switch.

FIG. 12 is a block diagram showing still another embodiment of the present invention. This embodiment is an embodiment applied to a hands-free car telephone.

12, the same symbols as in FIG. 11 indicate the same things. Reference numeral 25 is a hands-free communication circuit, 26 is a receiving speaker in place of the handset, 27 is a transmitting microphone in place of the transmitter, and 28 is a touch switch. The hand-free call circuit 25 is a circuit that enables a call without having a handset. The operator can talk by using the receiving speaker 26 and the transmitting microphone 27 without a handset. The touch switch 28 is a switch for transmitting a call intention or a call intention, and "ON" by one touch.
It becomes a structure that becomes "OFF" when it is touched once more.

Since this telephone does not have a handset, hook detection cannot be performed by lifting the handset as shown in FIG. 1, and this is replaced by the touch on the touch switch 28. The transmission microphone 27 is attached to a position where a voice of the operator can be collected with sufficient sensitivity, for example, a tie or the like. The touch switch 28 is also installed at a position where the operator can easily touch it, for example, at the same place as the transmission microphone 27. The receiving speaker 26 is installed at a position where the operator can listen at a required volume, such as a dashboard of a car.

FIG. 13 is an operation flow chart of this embodiment. Instead of the hook detection in the operation flow of FIG. 2, the operator touches the touch switch 28, and the operation state is switched by detecting that the switch is in the ON state. Basically, FIG. It is the same as that of. When the call ends, the operator touches the touch switch 28 again to turn it off.

FIG. 14 is a flowchart of the registration operation of the embodiment shown in FIG. Basically, it is the same as that of FIG. 3 showing the registration operation of the embodiment of FIG. 1, but the first proximity switch monitoring is replaced with the state monitoring of the touch switch 28. If the registration switch is pressed and the touch switch is in the ON state, repeated registration is possible. This is also
It can be modified in the same manner as that of FIG.

The call operation flow is the same as the call operation flow chart of the embodiment of FIG. 1 except that the first proximity switch monitoring in FIG. 5 or FIG. 7 is removed, and the touch switch 28 is turned on. When the input notification is completed, the voice recognition circuit 13 enters a standby state for the name utterance, and operates so as to continue until the specified time.

As described above, in this embodiment, the handset lifting and the detection of the proximity switch are replaced with the touch of the touch switch 28 by the operator. The basic operation is the same as that shown in FIG.

〔The invention's effect〕

According to the present invention, careless ambient noise is provided because the length of the voice input time to the voice recognition circuit is defined by the statistical property indicated by the time length (voice length) of the registered voice registered in advance in the voice recognition circuit. It is less likely that a malfunction will occur when such as.

Further, the start time of the voice input time is set to the time when the mouth is brought close to the transmitter (the time when the ear is brought close to the receiver) by using the output of the proximity switch arranged near the handset transmitter. This prevents the mechanical sound generated when the handset is lifted from being input to the voice recognition circuit and eliminates the malfunction.

Further, by starting the voice input time after a certain period of time has passed since the handset was lifted, it is possible to prevent a mechanical sound caused by lifting the handset from becoming an input voice to the voice recognition circuit and eliminate a malfunction. According to the present invention, the above effects can be expected.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIGS. 2 to 7 are flow charts showing the flow of its operation, and FIG.
FIG. 11 is a block diagram showing another embodiment of the present invention, FIG. 9 and FIG. 10 are flow charts showing the flow of its operation, respectively.
FIG. 12 and FIG. 12 are block diagrams showing still another embodiment of the present invention, and FIG. 13 and FIG. 14 are flow charts showing the flow of the operation. Explanation of symbols 7 ... Proximity switch, 13 ... Voice recognition circuit, 14 ...
Voice length measuring circuit, 15 ... Voice length memory, 16 ... Voice length statistical circuit, 17 ... Regular time register, 20 ... Hook detection circuit, 22 ... Timer circuit, 24 ... Mobile unit,
25: Hand-free call circuit, 28: Touch switch

Claims (8)

[Claims] 1. A voice recognition means for registering a voice to be recognized in advance and recognizing the input voice by comparing the input voice with a registered voice when the voice is inputted from the voice input means.
A voice dial device comprising a dial signal transmitting means for transmitting a dial signal to a telephone line based on the recognition result, and a voice length measuring means for measuring the length (voice length) of the registered voice in advance, A voice length statistical means for performing statistical processing on the entire registered voice length and calculating a certain prescribed time as a result, a prescribed time registration means for registering the prescribed time, and a voice input means for the voice recognition means. A voice dialing device, comprising: a voice input acceptance limiting means for limiting a voice input acceptance time via the specified time. 2. A voice dial device according to claim 1, wherein proximity detecting means for detecting the proximity of a speaker's ear or mouth is provided near the voice input means. A voice dial device, characterized in that a detection output time of the proximity detection means is set to a start time of the voice input reception time. 3. The voice dial device according to claim 1, further comprising an input reminder means for prompting a speaker as a dialer to input a voice, and a speaker near the voice input means. Proximity detection means for detecting the proximity of the ear or mouth of the person is provided, and the input notification means is operated by the detection output of the proximity detection means to prompt the speaker to input a voice and the voice recognition. A voice dial device, characterized in that the voice reception to the means is started. 4. The voice dial device according to claim 1, wherein the voice input means is a transmitter with a built-in handset, and the voice input reception time is reached when a certain time has elapsed after the handset was off-hook. A voice dial device characterized in that it is the start point of. 5. A voice dial device according to claim 1, further comprising an input reminder means for prompting a speaker as a dialer to input a voice, and the voice input means has a built-in handset. And activating the input notification means by an output signal indicating that a certain time has elapsed after the handset is off-hook, prompting the speaker to input a voice, and starting voice reception to the voice recognition means. A voice dial device characterized by. 6. The voice dial device according to claim 1, further comprising a touch switch for allowing a speaker as a dialer to convey a willingness to make a call, and the touch switch is touched when the switch is touched. A voice dial device characterized in that it is the start time of a voice input reception time. 7. A voice dial device according to claim 1, wherein when a voice input is detected within the specified time, the measuring means for the specified time is once cleared at the time of the detection and the input is repeated. A voice dial device that starts measuring a specified time. 8. The voice dial device according to claim 2 or 3, wherein the voice input means is a handset in a handset.