JP2837577B2 - Answering machine - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an answering machine having a voice recognition function, a speaker recognition function, and a voice analysis / synthesis function.

[0002]

2. Description of the Related Art Conventionally, there is an answering machine as shown in a block diagram of FIG. This answering machine operates as follows. In other words, "answering machine mode" before going out
In order to prepare for the above, the user operates a predetermined key row of the key input unit 25 to instruct the control unit 24 to enter an “outgoing message (OGM) registration mode”.

[0003] Then, the voice analysis / synthesis unit 21 performs the following voice synthesis under the control of the control unit 24. That is, the voice analysis / synthesis unit 21 itself analyzes and reads the guidance voice parameters stored in the synthesis memory 22, converts the guidance voice parameters into a voice waveform, and sends the voice waveform to the receiver 27. Thus, the guidance based on the synthesized voice is output from the receiver 27. After that, the voice analysis / synthesis unit 21 performs the following voice analysis under the control of the control unit 24. That is, the user's voice waveform input from the microphone 26 is read and analyzed in accordance with the guidance based on the synthesized voice, and the OGM voice data obtained by compression is stored in the O / M
The data is stored in a GM area (not shown) and the “OGM registration mode” ends.

When the telephone number of the answering machine is received via the telephone line 29 while the apparatus is in the "answering machine mode" and on standby, the control section 24 causes the line connecting section 28 to
Is controlled to establish connection with the telephone line 29. Then,
The voice analysis / synthesis unit 21 performs the following voice synthesis under the control of the control unit 24. That is, the OGM audio data stored as described above is read from the OGM area in the recording / playback memory 23, converted into an audio waveform, and transmitted to the line connection unit 28. In this way, the OGM recorded in advance by the user is reproduced as an audio signal and transmitted to the telephone line 29.

After that, the speech analysis / synthesis unit 21
When the voice waveform of a message (ICM) from the caller transmitted from the telephone line 29 according to the OGM based on the reproduced sound is received via the line connection unit 28, the voice waveform of the caller is read. The ICM voice data obtained by analyzing and compressing the data is stored in the ICM memory
Store in area (not shown).

The user who has returned from the house operates a predetermined key sequence from the key input unit 25 to listen to the ICM received while he / she is away, and instructs the control unit 24 to end the above-mentioned "answering machine mode". Instruct the start of the “ICM playback mode”. Then, the voice analysis / synthesis unit 21 controls the control unit 24
Under the control of, the ICM audio data is read from the ICM area in the recording / playback memory 23, reproduced as an audio signal, and transmitted to the receiver 27. Thus, the ICM from the caller is output from the receiver 27 by the reproduced sound.

As described above, in a conventional answering machine,
At the time of an incoming call, based on the OGM audio data read from the OGM area in the recording / playback memory 23,
OGM based on the reproduced sound is transmitted to the caller. Therefore, unless the user updates the OGM voice data, the same response is made to all incoming calls, and there is a disadvantage that an appropriate response cannot be transmitted according to the caller. Further, since the ICM voice data is simply stored in the recording / playback memory 23 as it is, when the user knows the ICM addressed to himself, all ICMs are stored.
There is also a disadvantage that the contents of M must be determined by listening to the reproduced sound.

In order to solve the above-mentioned drawbacks, the following answering machine has been proposed. JP-A-61-
In Japanese Patent No. 39756, a plurality of previously recorded OGs are recorded by incorporating a voice recognition device in an answering machine.
One of M is selected according to the recognition result of the keyword uttered by the caller and transmitted to the telephone line. In this way, it is possible to inform a specific person who wants to contact immediately of his / her own location and the nearest telephone number. In Japanese Patent Application Laid-Open No. 61-43862, a voice recognition device is similarly incorporated in an answering machine, and based on a standard pattern of a keyword of a specific speaker registered in advance, a speech of a keyword spoken by a caller is spoken. Perform person recognition. In this way, the caller can automatically notify the user whether or not the user is the specific party desired.

In Japanese Patent Application Laid-Open No. 63-67949, a voice recognition device is incorporated in an answering machine to recognize a specific called party name (user name) spoken by a calling party. The corresponding called party information is stored in the storage unit in association with the ICM. In this way, when a single answering machine is shared by a plurality of users, it is possible to easily identify which user has called the telephone by reproducing the called party information stored in the storage unit. .

[0010]

However, the answering machine incorporating the above-mentioned conventional voice recognition device has the following problems. That is, in the case of JP-A-61-39756 and JP-A-61-43862, a cumbersome registration operation in which a specific speaker is made to utter a keyword in advance and a standard pattern of the specific speaker is extracted and registered. There is a problem that requires. Further, since only information on who the caller is is obtained by voice recognition, there is also a problem that, when there are a plurality of users, OGM cannot be transmitted according to the combination of the caller and the called party. .

In the case of Japanese Patent Application Laid-Open No. 63-67949, called party information such as a called party name stored in a storage unit in association with the ICM is reproduced, and the called party information is reproduced. Is the called party information related to the user, the ICM associated with the called party information is output as a synthesized sound by key input designation. Therefore, there is a problem that the operation until the user who has returned from home to obtain the ICM for himself is troublesome. Also, based on the recognition result of the called party name, only called party information such as the called party name can be obtained, and when there are a plurality of calling parties, the combination of a specific calling party and the called party is determined. There is also a problem that only the corresponding ICM cannot be reproduced.

An object of the present invention is to provide an answering machine which can easily and automatically record / transmit OGM from a specific user to a specific party via a telephone line, or facilitate ICM from a specific party to a specific user. To provide an answering machine that can automatically record and play back.

[0013]

In order to achieve the above object, the answering machine of the first invention analyzes a voice of a name of a user or a registrant inputted at the time of registration and a word of a user name or a registrant name. A standard pattern is obtained and a user number or a registrant number is added and registered in the word pattern memory. A word recognition unit for obtaining a user number or a registrant number added to a word standard pattern, and analyzing the voice of the input user or registrant during registration to obtain a speaker standard pattern and obtaining the user number or the registration. While registering a speaker number and adding it to the speaker pattern memory, the sound generated by the user or registrant input during recognition is And a speaker recognition unit for obtaining a user number or a registrant number added to a speaker standard pattern obtained by recognizing the OGM voice from a specific user to a specific registrant. A voice analysis / synthesis unit that adds the corresponding user number and registrant number to the obtained OGM voice data and stores the data in a recording / playback memory;
During recording, the word recognition unit, speaker recognition unit, and voice analysis /
A control unit for controlling the synthesizing unit to add a user number and a registrant number corresponding to the OGM audio data of the OGM from the specific user to the specific registrant based on the voice of the specific user and record the added data It is characterized by:

The answering machine of the second invention is a telephone answering machine,
In the answering machine according to the invention, the voice analysis / synthesis unit reads out specific OGM voice data from the recording / playback memory based on the input user number and registrant number during playback, and plays back the voice signal. The control unit controls the word recognition unit, the speaker recognition unit and the voice analysis / synthesis unit by automatically responding to an incoming call from the telephone line in the answering machine standby state, and The OGM voice data to which the user number and the registrant number obtained based on the voice of the caller are added is read out, reproduced as a voice signal, and transmitted to the telephone line. .

The answering machine of the third invention analyzes the voice of the input name of the user or the registrant at the time of registration, obtains a word standard pattern of the user name or the registrant name, and obtains the user number or the registrant number. Is added to the word pattern memory, and at the time of recognition, the user number obtained by recognizing the voice of the input name of the input user or registrant or the user number or the user number added to the word standard pattern of the registrant name A word recognition unit for obtaining a registrant number, and at the time of registration, a voice generated by the input of a user or a registrant is analyzed to obtain a speaker standard pattern, and the user number or the registrant number is added to the speaker pattern memory. While registering, the speaker standard pattern obtained by recognizing the voice of the input user or registrant at the time of recognition And a speaker recognizing unit for obtaining a user number or a registrant number added to the registrant, and a registrant corresponding to the ICM voice data obtained by analyzing and compressing the ICM voice input from the specific registrant to the specific user. A voice analysis / synthesis unit for adding a number and a user number to the recording / playback memory, and automatically responding to an incoming call from the telephone line in an answering machine standby state, the word recognition unit, the speaker recognition unit, and the voice analysis unit. Controlling the synthesizing unit, and based on the caller's voice input from the telephone line, the caller's registrant number and the ICM voice data of the ICM from the caller to the called party. The recording apparatus further comprises a control unit for adding the called party's user number and recording.

The answering machine of the fourth invention is a telephone answering machine of the third invention.
In the answering machine of the invention, the voice analyzing / synthesizing unit reads out specific ICM voice data from the recording / playback memory based on the input registrant number and user number during playback and plays back the voice signal. The control unit controls the word recognition unit, the speaker recognition unit, and the voice analysis / synthesis unit during the reproduction after the answering machine is on standby, and the registrant number obtained based on the voice of the specific user. And the specific ICM voice data to which the user number and the user ID are added is read out, reproduced as a voice signal, and transmitted to the receiver.

Also, the answering machine according to the fifth invention is characterized in that
In the answering machine according to the invention, the voice analysis / synthesis unit analyzes and compresses the input ICM voice from the specific registrant to the specific user and converts the ICM voice data to a corresponding registrant number and user number. Is stored in the recording / playback memory, and at the time of playback after the answering machine standby, the specific ICM voice data is read out from the recording / playback memory based on the input registrant number and user number. The control unit, when automatically responding in the answering machine standby state,
Registration of the caller in ICM voice data of the ICM from the caller to the called party which is input in response to the OGM transmitted to the telephone line based on the input caller's voice In addition to the recording of the caller's number and the called party's user number, during reproduction after the answering machine standby, the specific ICM to which the registrant number and user number obtained based on the utterance of the specific user are added. The audio data is read out, reproduced into an audio signal, and transmitted to a receiver.

[0018]

According to the first aspect of the present invention, the word recognition unit preliminarily registers a user standard or a registrant name by adding a user number or a registrant number to the word standard pattern of the user name and the registrant name. In the speaker pattern memory, a speaker number is added to a speaker standard pattern of a user and a registrant by a speaker recognition unit and registered.

Thereafter, at the time of recording, the word recognition unit and the speaker recognition unit are controlled by the control unit, and the user number of the specific user and the registrant number of the specific registrant are obtained based on the voice of the specific user. Can be Also, voice analysis
The synthesizing unit is controlled so that the input OGM voice from the specific user to the specific registrant is analyzed and compressed, and the obtained user number and registrant number are added to the OGM voice data obtained. Then, it is stored in the recording / playback memory.

Thus, the OGM from the specific user to the specific registrant is recorded in association with the user number of the specific user and the registrant number of the specific registrant.

Further, in the second invention, when an OGM from a specific user to a specific registrant is recorded by the voice analysis / synthesis unit and the telephone enters the answering machine standby state and there is an incoming call from the telephone line, the control unit causes the next operation. An automatic response such as is performed. That is, the word recognition unit and the speaker recognition unit are controlled, and the user number of the called party and the registrant number of the calling party are obtained based on the voice of the calling party from the telephone line. On the other hand, the voice analysis / synthesis unit is controlled, and based on the obtained user number and registrant number, a specific OGM for which the user number and the registrant number are added from the recording / playback memory. The audio data is read out, reproduced as an audio signal, and transmitted to the telephone line.

In this way, even if various OGMs assigned from a plurality of users to a plurality of registrants are stored in the recording / playback memory, a specific user (called party) can be sent to a specific registrant (calling party). The OGM is automatically selected and reliably sent to the caller.

In the third invention, when an incoming call is received in the answering machine standby state, the control unit controls the word recognition unit and the speaker recognition unit, and the control unit controls the word recognition unit and the speaker recognition unit based on the voice of the caller from the telephone line. The calling party's registrant number and the called party's user number are determined. On the other hand,
The synthesizing unit is controlled so that the caller (specific registrant) input from the telephone line to the called party (specific user) receives
The obtained registrant number of the calling party and the user number of the called party are added to the ICM voice data obtained by analyzing and compressing the CM voice and stored in the recording / playback memory. Is done.

Thus, the ICM from the specific registrant to the specific user is recorded in association with the registrant number of the specific registrant and the user number of the specific user.

In the fourth invention, the ICM from the calling party to the called party is recorded by the voice analysis / synthesis unit when an incoming call arrives in the answering machine standby state, and the answering machine standby ends. After that, when the ICM is reproduced by the specific user who has returned home, first, the word recognition unit and the speaker recognition unit are controlled by the control unit, and based on the voice of the specific user, the registrant of the specific registrant is registered. A number and a user number of the specific user are determined. Next,
The synthesizing unit is controlled so that the specific ICM audio data to which the obtained registrant number and user number are added is read out from the recording / playback memory, reproduced as an audio signal, and output from the receiver. .

In this way, even if various ICMs assigned to a plurality of users from a plurality of registrants are stored in the recording / reproducing memory, a specific registrant (caller) can send a message to a specific user (called party). The ICM is automatically selected irrespective of the designation by key input or the like and is reliably transmitted to the specific user.

In the fifth invention, after an OGM from a specific user to a specific registrant is recorded by the voice analysis / synthesis unit and enters an answering machine standby state, when an incoming call is received from a telephone line, the called party is called. The specific OGM to which the user number and the registrant number of the caller are added is read out by the voice analysis / synthesis unit and transmitted to the telephone line.

Further, when an ICM applied to the called party from the calling party is input via the telephone line in response to the OGM, the control section controls the voice analysis / synthesizing section, The input ICM voice is analyzed and compressed to obtain ICM voice data. Then, the registrant number and the user number obtained based on the voice of the caller by controlling the word recognition unit and the speaker recognition unit are added to the obtained ICM voice data and recorded. Stored in the playback memory. Thus, the answering machine wait ends.

After that, when the ICM is reproduced by the specific user who has returned home, first, the word recognition unit and the speaker recognition unit are controlled by the control unit, and based on the voice of the specific user, the specific registrant is registered. Of the registrant and the user number of the specific user. Next, the voice analysis / synthesis unit is controlled so that the specific ICM voice data to which the obtained registrant number and user number are added is read from the recording / playback memory and played back as a voice signal. Output from the receiver.

In this way, even if various OGMs applied from a plurality of users to a plurality of registrants are recorded in the recording / playback memory, a callee (specific user) to a caller (specific registrant) can be transmitted. The OGM is automatically selected and reliably transmitted to the specific registrant. Further, at the time of ICM reproduction, even if various ICMs applied from a plurality of registrants to a plurality of users are recorded in the recording / playback memory, the ICM from the specific registrant to the specific user is input by key input or the like. It is automatically selected regardless of the designation and is reliably transmitted to the specific user.

[0031]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the illustrated embodiments. FIG. 1 is a block diagram of the answering machine of the present embodiment. This answering machine has a word recognition section 1, a word pattern memory 2, a voice analysis / synthesis section 3, an echo back sound memory 4, a recording / playback memory 6, a speaker recognition section 8, a speaker pattern memory 9, and a line connection section 13. And a control unit 7 and the like.

The word recognizing section 1 stores a user name (a user's name, which corresponds to the called name of the called party at the time of an incoming call).
Alternatively, when a voice signal of the registrant name (name of the other party with which the user talks and corresponds to the name of the caller at the time of the incoming call) is input from the microphone 11, the input voice signal is analyzed and word unit is analyzed. To obtain a time series of the characteristic parameters of Then, the time series of the characteristic parameters is stored in the word pattern memory 2 as a word standard pattern. At this time, the word standard pattern of the user name is stored by adding a user number (called number at the time of an incoming call) representing a user to a user name pattern area (not shown) in the word pattern memory 2. On the other hand, the word standard pattern of the registrant name is stored by adding a registrant number (caller number at the time of an incoming call) indicating a registrant to a registrant name pattern area (not shown) in the word pattern memory 2. You do it.

Further, a characteristic pattern consisting of the characteristic parameter time series obtained by analyzing the voice signal input from the telephone line 14 via the line connection unit 13 and the word pattern memory 2 It performs matching with a plurality of word standard patterns stored in the user name pattern area in the. And the user number of the word standard pattern obtained as a result of recognition by matching
(The called party number at the time of the incoming call) is sent to the control unit 7. On the other hand, the feature pattern consisting of the feature parameter time series obtained by analyzing the voice signal input from the microphone 11 and the feature pattern previously stored in the registrant name pattern area in the word pattern memory 2 as described above. Matching with multiple word standard patterns. Then, the registrant number of the word standard pattern (the caller number at the time of the incoming call) obtained as a result of the recognition by the matching is transmitted to the control unit 7.

When a voice signal of a user name or a registrant name is input from the microphone 11, the voice analysis / synthesis unit 3 analyzes and compresses the waveform of the input voice signal, and converts the obtained parameters. The data is stored in the echo-back sound memory 4 as the sound data for the echo-back sound. At that time,
The echo back sound voice data of the user name is stored in a user name area (not shown) in the echo back sound memory 4, while the echo back sound voice data of the registrant name is stored in the echo back sound memory 4. It is stored in the registrant name area (not shown).

Further, when an OGM audio signal from a specific user to a specific registrant is input from the microphone 11, the waveform of the input audio signal is analyzed and compressed.
The obtained parameters are used as OGM audio data in an OGM area (not shown) in the recording / playback memory 6 and the user number of the specific user (the called party number at the time of the incoming call) and the registrant number of the specific registrant ( (Caller number at the time of incoming call). On the other hand, when an ICM voice signal from a specific registrant to a specific user is input via the telephone line 14, the waveform of the input voice signal is analyzed and compressed, and the obtained parameters are converted into ICM voice data. ICM area in the recording / playback memory 6
A pair of the registrant number of the specific registrant (caller number at the time of the incoming call) and the user number of the specific user (the called number at the time of the incoming call) is added to (not shown) and stored.

The voice analysis / synthesis unit 3 reads out the voice data for the echo back sound stored in the user name area or the registrant name area in the echo back sound memory 4 and generates a voice signal. The echo back sound signal thus generated is sent to the line connection unit 13 or the receiver 12. Further, the voice synthesizing parameter stored in the synthesizing memory 5 is read out to generate a voice signal. The voice signal of the synthesized voice guidance thus generated is sent to the line connection unit 13 or the receiver 12.

The voice analysis / synthesis unit 3 generates a pair of the user number and the registrant number from the OGM area in the recording / playback memory 6 based on the input pair of the user number and the registrant number. The OGM audio data to which is added is read out, reproduced as an audio signal, and transmitted to the line connection unit 13. On the other hand, based on the pair of the registrant number and the user number, the ICM area in which the pair of the registrant number and the user number are added from the ICM area in the recording / playback memory 6.
The audio data for use is read out, reproduced as an audio signal, and transmitted to the receiver 12.

When a speech signal of an arbitrary sentence uttered by the user is input from the microphone 11, the speaker recognition unit 8 analyzes the input speech signal and analyzes the input speech signal for a word-based feature parameter (power spectrum). And long term average of pitch frequency). Then, the pattern of the characteristic parameter is set as a speaker standard pattern, and the above-mentioned user number (called number at the time of an incoming call) is added thereto and stored in a user speaker pattern area (not shown) in the speaker pattern memory 9. . On the other hand, when a voice signal from the registrant is input from the line connection unit 13, the input voice signal is analyzed to obtain a word-based characteristic parameter (power spectrum, long-term average of pitch frequency, etc.). Then, in accordance with the registration instruction from the key input unit 10, the obtained characteristic parameter is used as a speaker standard pattern, and the above-mentioned registrant number (caller number at the time of an incoming call) is added and registered in the speaker pattern memory 9. It is stored in a speaker pattern area (not shown).

Further, a speaker characteristic pattern obtained by analyzing the voice signal input from the microphone 11
As described above, matching is performed with the standard speaker patterns of a plurality of users stored in the user speaker pattern area in the speaker pattern memory 9 in advance. Then, the control unit 7 sends the resulting user number (called number) to the user's speaker standard pattern. On the other hand, the speaker characteristic pattern obtained by analyzing the voice signal input from the line connection unit 13 and the speaker characteristic pattern previously stored in the registered speaker pattern area in the speaker pattern memory 9 as described above. Matching with speaker standard patterns of a plurality of registrants. Then, the registrant number (caller number) added to the speaker standard pattern of the registrant obtained as a result is sent to the control unit 7.

At the time of transmission, when a signal indicating the telephone number of the other party is input from the control unit 7 during transmission, the line connection unit 13 transmits the telephone number to the telephone line 14. After the telephone line of the other party is connected to the telephone line, the voice signal input from the microphone 11 or the voice analysis / synthesis unit 3 is transmitted to the telephone line 14. At the time of reception, the voice signal received from the telephone line 14 is converted under the control of the control unit 7 into the word recognition unit 1, the speaker recognition unit 8, the voice analysis /
The signal is sent to the combining unit 3 or the receiver 12.

The control unit 7 controls the word recognition unit 1, the voice analysis / synthesis unit 3, the speaker recognition unit 8, the line connection unit 13, and the like to recognize the voice from the user or the registrant. The registrant (the caller at the time of the incoming call) and the user (the called party in the caller) are specified. Then, the OGM corresponding to the pair of the specific user and the specific registrant is transmitted to the telephone line, and the ICM from the specific registrant to the specific user is recorded / correlated with the pair of the specific registrant and the specific user. To play. Further, an instruction or the like to the control unit 7 is input from the key input unit 10.

The answering machine having the above configuration operates as follows. There are three operation modes of the answering machine: a registration mode for registering the word standard pattern and the speaker standard pattern in advance, a recognition mode for recognizing the input voice, and a recording / reproduction mode for recording / reproducing OGM and ICM. There is a mode. Control of switching between these three operation modes and other operation modes described in detail later is performed by the control 7. Hereinafter, each operation mode will be described in detail.

<Registration Mode> This registration mode is further divided into five operation modes: a user name registration mode, a user pattern registration mode, a registrant name registration mode, a registrant pattern analysis mode, and a registrant pattern registration mode.

[User Name Registration Mode] This operation mode is an operation mode in which a user registers his / her own name in advance. In other words, the user uses the speech analysis / synthesis unit 3 based on the speech synthesis parameters stored in the synthesis memory 5.
According to the instruction of the voice guidance by the synthesized voice generated by the user, his / her own name (that is, the user name) is uttered toward the microphone 11. Then, the voice signal output from the microphone 11 is input to the word recognition unit 1 and the voice analysis / synthesis unit 3.

Then, the word recognizing unit 1 adds the user number (called number) to the word standard pattern of the user name obtained as described above and stores the user name pattern in the word pattern memory 2. Register in the area.
At the same time, the voice analysis / synthesis unit 3 stores the voice data for the echo-back sound obtained as described above in the user name area in the echo-back sound memory 4. At this time, the control unit 7 stores link information between the word standard pattern stored in the word pattern memory 2 and the echo back sound parameter stored in the echo back sound memory 4 in an internal memory (not shown). Store.

[User Pattern Registration Mode] This operation mode is an operation mode in which the user registers the characteristics of his or her own voice in advance. That is, the user performs 1 according to the voice guidance based on the synthesized voice generated as described above.
An arbitrary sentence of about 0 seconds is uttered toward the microphone 11. Then, the audio signal output from the microphone 11 is input to the speaker recognition unit 8. Then, in the speaker recognition section 8, the user number (called number) is added to the user standard pattern obtained as described above.
Is stored in the user speaker pattern area in the speaker pattern memory 9.

[Registrant Name Registration Mode] This operation mode is an operation mode in which the user registers the name of a partner who transmits OGM or receives ICM. That is, the user speaks the microphone (name of the registrant) toward the microphone 11 according to the voice guidance as described above. And
The voice signal output from the microphone 11 is input to the word recognition unit 1 and the voice analysis / synthesis unit 3.

Then, the word recognition section 1 adds the registrant number (caller number) to the word standard pattern of the registrant name obtained as described above, and Register in the registrant name pattern area. At the same time, the voice analysis / synthesis unit 3 stores the echo back sound parameter obtained as described above in the registrant name area in the echo back sound memory 4.
At this time, the control unit 7 internally stores link information between the registrant's word standard pattern stored in the word standard pattern memory 2 and the registrant's echo back sound parameter stored in the echo back sound memory 4. Store in memory.

[Registrant pattern analysis mode] This operation mode is used when the answering machine is connected to the telephone line of the caller (registrant) and a line is being carried out between the user and the registrant. And an operation mode performed when an ICM is transmitted from a caller. That is, the voice signal of the registrant transmitted via the telephone line 14 is input from the line connection unit 13 to the speaker recognition unit 8. Then, the speaker recognition unit 8 calculates the speaker standard pattern of the registrant (that is, the long-term (talk time) average value such as the power spectrum and the pitch frequency) as described above. The registrant speaker standard pattern thus calculated is temporarily stored in the buffer.

[Registerer Pattern Registration Mode] This operation mode is an operation mode in which the speaker standard pattern obtained by the above-mentioned "registrant pattern analysis mode" is formally registered. That is, after the call with the registrant ends or
When the registration of the speaker standard pattern is instructed by key input from the key input unit 10 after the CM is reproduced, the registrant number (caller number) is added to the registrant speaker standard pattern stored in the buffer. ) Is added to the speaker pattern memory 9
In the registrant speaker pattern area.

<Recognition Mode> This recognition mode is further divided into four operation modes: a user name recognition mode, a user recognition mode, a registrant name recognition mode, and a registrant recognition mode.

[User Name Recognition Mode] This operation mode is an operation mode for recognizing a called party name (ie, a user name) at the time of an incoming call. That is, a voice signal indicating a user name by voice of the caller transmitted via the telephone line 14 is input from the line connection unit 13 to the word recognition unit 1. Then, the word recognition unit 1 sends the user number (called number) of the word standard pattern obtained as a result of performing the matching and recognizing the input voice to the control unit 7.

Then, the user number (called number)
The control unit 7 having received the voice control unit 7 controls the voice analysis / synthesis unit 3 to refer to the link information stored in the internal memory and to store the corresponding echo-back sound voice data in the user in the echo-back sound memory 4. Read from the name area. Then, the voice analysis / synthesis unit 3 generates a voice signal indicating the user name based on the voice data for the echo back sound, and sends it to the line connection unit 13. Thus, confirmation of the recognition result is prompted by the echo back sound.

[User Recognition Mode] In this operation mode,
This is an operation mode for recognizing the user at the time of OGM recording and ICM reproduction, which will be described in detail later. In other words, the OGM output from the microphone 11 by the user's utterance or an audio signal of an arbitrary sentence is input to the speaker recognition unit 8.
Then, the speaker recognition unit 8 sends the user number (called number) of the speaker standard pattern obtained as a result of performing the above-described matching and recognizing the input voice to the control unit 7. However, at the time of the above matching, if the distance from any of the speaker standard patterns is longer than the threshold value, the user number (called number) “0” is transmitted.

[Registrant Name Recognition Mode] This operation mode is an operation mode for recognizing a registrant name at the time of OGM recording or ICM reproduction, which will be described in detail later. That is,
A voice signal of the registrant name output from the microphone 11 and uttered by the user is input to the word recognition unit 1. Then, the word recognition unit 1 sends the registrant number (caller number) of the word standard pattern obtained as a result of performing the above-described matching and recognizing the input voice to the control unit 7.

Then, the registrant number (caller number)
The control unit 7 that has received the data stores the corresponding echo-back sound voice data in the echo-back sound memory 4 with reference to the link information stored in the internal memory by controlling the voice analysis / synthesis unit 3. Read from the user name area. Then, the audio analysis / synthesis unit 3 generates an audio signal based on the echo-back sound audio data, and sends it to the receiver 12.

[Registrant Recognition Mode] This operation mode is
This is an operation mode for recognizing a caller (that is, a specific registrant) at the time of an incoming call. That is, the voice signal of the caller transmitted via the telephone line 14 is input from the line connection unit 13 to the speaker recognition unit 8. Then, the speaker recognizing section 8 sends the registered number (caller number) of the speaker standard pattern obtained as a result of performing the above-mentioned matching and recognizing the input voice to the control section 7. However, at the time of the above matching, if the distance from any of the speaker standard patterns is longer than the threshold value, the registrant number (caller number) “0” is transmitted.

<Recording / Playback Mode> This recording / playback mode is further divided into three operation modes: an OGM recording mode, an ICM recording mode, and an ICM playback mode.

[OGM recording mode]
This is an operation mode in which the OGM for the registrant is recorded when the user goes out. That is, the OGM audio signal from the user output from the microphone 11 is input to the audio analysis / synthesis unit 3. Then, the audio analysis / synthesis unit 3 analyzes the audio waveform and stores the OGM audio data obtained by compression in an OGM area (not shown) in the recording / playback memory 6. At this time, a pair of the user number (called number) of the user who transmits the OGM voice data and the registrant number (calling number) of the registrant receiving the OGM voice data is stored in the OGM.
It is added to the audio data for use and stored.

[ICM recording mode]
This is an operation mode in which an ICM from a registrant is recorded while the user is out. That is, the voice signal of the ICM transmitted by the registrant transmitted via the telephone line 3 is input from the line connection unit 13 to the voice analysis / synthesis unit 3. Then, the voice analysis / synthesis unit 3 analyzes the voice waveform and stores the ICM voice data obtained by compression in an ICM area (not shown) in the recording / playback memory 6. At this time, this I
The pair of the registrant number (caller number) of the registrant who originated the CM and the user number (called number) of the receiving user is represented by the I
It is added to the CM and stored.

[ICM playback mode]
This is an operation mode for reproducing the ICM from the registrant when the user returns from home. That is, the voice analysis / synthesis unit 3
Recording and playback of an ICM to which a pair of a user number (called number) of a user who plays the ICM and a registrant number (calling number) of a caller (registrant) of the target ICM is added. The data is read from the ICM area in the memory 6 for use, converted into an audio waveform, and transmitted to the receiver 12.

[0062] The control unit 7, as described in detail below,
The word recognition unit 1, the voice analysis / synthesis unit 3, and the speaker recognition unit 8 are controlled to switch the above-described operation modes,
“Initial registration processing”, “Call registration processing”, “OGM recording processing”,
The respective processes of “incoming process” and “ICM playback process” are executed. FIG. 2 is a flowchart of an operation mode performed when each of the above processes is executed. Hereinafter, according to FIG.
An operation mode performed in each process will be described.

Normally, the control unit 7 is in a standby state,
It waits for designation of the execution process from the key input unit 10. If "initial registration process" is designated in this state, "user name registration mode" and "user pattern registration mode" are sequentially performed, and the word standard pattern of the user name and the speaker standard pattern of the user are registered. You. When the “initial registration process” ends, the process returns to the standby state.

When a call is started by being connected to the answering machine and the telephone line, a "call registration process" is started, and a "registrant pattern analysis mode" is first executed to call the other party.
The (registrant) speaker standard pattern is calculated. Subsequently, if there is a registration instruction from the key input unit 10, the "registrant name registration mode" and the "registrant pattern registration mode" are performed, and the word standard pattern of the call name (registrant name) of the call partner and the call partner are registered. The (registrant) speaker standard pattern is registered.
When the “call registration process” is completed in this way, the process returns to the standby state.

When "OGM recording process" is designated, "registrant name recognition mode", "OGM recording mode" and "user recognition mode" are sequentially executed, and the OGM audio data is stored in the user number (received). A pair of a caller number) and a registrant number (caller number) is added and stored. Thus, "O
When the "GM recording process" ends, the process returns to the standby state.

When there is an incoming call when the standby state is the answering machine standby state, "incoming processing" is entered, and "user name recognition mode", "registrant recognition mode", "ICM recording mode" and "registrant Pattern analysis mode "is sequentially performed. Then, the ICM voice data is stored with the registrant number (calling party number) and the user number (called party number) added thereto, and the speaker standard pattern of the calling party is calculated. When the "incoming process" is completed, the process returns to the standby state.

When "ICM playback processing" is designated, "user recognition mode", "registrant name recognition mode" and "ICM playback mode" are sequentially executed. as a result,
ICM voice data to which the recognized registrant number (caller number) and user number (callee number) are added is read out and output from the receiver 12. Subsequently, if there is a registration instruction from the key input unit 10, the "registrant name registration mode" and the "registrant pattern registration mode" are executed, and the word standard pattern and the speaker standard pattern of the registrant name are additionally registered. . When the “ICM reproduction process” is completed in this way, the process returns to the standby state.

Hereinafter, each of the above processes will be described in further detail in accordance with a case where the answering machine is actually used. However, a detailed description of the routine of each operation mode is omitted.
First, prior to using the answering machine, an initial registration process for registering information on each user using the answering machine is performed.

FIG. 3 is a flowchart of the initial registration operation performed by the word recognition unit 1, the voice analysis / synthesis unit 3, and the speaker recognition unit 8 under the control of the control unit 7. Hereinafter, the initial registration processing operation will be described in detail with reference to FIG. In step S1, it is determined whether the "registration key" in the key input unit 10 has been pressed. As a result, if pressed, the process proceeds to step S2 to start the initial registration processing operation. In step S 2, the voice analysis / synthesis unit 3 generates the synthesized voice guidance “Please say your name” and outputs it from the receiver 12.

In step S 3, the above-described “user name registration mode” routine is executed by the word recognition unit 1. As a result, the word standard pattern of the user name uttered by the user according to the synthetic voice guidance is
A user number (called number) is added to a user name putter area in the word pattern memory 2 and registered. In step S 4, the voice analysis / synthesis unit 3 generates the synthesized voice guidance “Please speak for about 10 seconds” and outputs it from the receiver 12. In step S5, the above-described "user pattern registration mode" routine is executed by the speaker recognition unit 8, and the initial registration processing operation ends.
Then, it enters a standby state. As a result, the speaker standard pattern extracted from the voice signal of an arbitrary sentence uttered by the user according to the synthesized voice guidance is stored in the user speaker putter area in the speaker pattern memory 9 with the user number.
(Called party number) is added and registered.

Subsequently, a call registration process is performed in which the user registers information on a partner who wants to transmit and receive OGM and ICM. FIG. 4 is a flowchart of a call registration processing operation performed by the word recognition unit 1, the voice analysis / synthesis unit 3, and the speaker recognition unit 8 under the control of the control unit 7.
Hereinafter, the call registration processing operation will be described in detail with reference to FIG. In step S11, it is determined whether or not the telephone line 14 has been connected to the answering machine. As a result, if connected, the process proceeds to step S12. In step S12, it is determined whether or not the call has started. As a result, if the call has been started, the process proceeds to step S13, and the call registration processing operation is started. In step S13, the above-described "registrant pattern analysis mode" routine is executed by the speaker recognition unit 8. As a result, the telephone line 14
The speaker standard pattern calculated from the voice signal of the other party (calling party) transmitted via the above is temporarily stored in the buffer.

In step S14, it is determined whether or not the telephone line 14 is still connected to the answering machine. As a result, if the connection is not established (that is, if the call is terminated), the process proceeds to step S15, and if not, the process waits for the call to be terminated. In step S15, it is determined whether the "registration key" in the key input unit 10 has been pressed.
As a result, if pressed, the process proceeds to step S16; otherwise, the call registration processing operation is terminated and a standby state is entered.

In step S16, the speech analysis / synthesis unit 3
As a result, the synthesized voice guidance “Please speak the other party's name” is generated and output from the receiver 12.
In step S17, the above-mentioned "registrant name registration mode" routine is executed by the word recognition unit 1. As a result, the word standard pattern of the registrant name uttered by the user according to the synthetic voice guidance is registered with the registrant number (caller number) added to the registrant name putter area in the word pattern memory 2. . In step S18,
The above-mentioned "registrant pattern registration mode" is executed by the speaker recognition unit 8, and the call registration processing operation is completed.
Then, it enters a standby state. As a result, the above step S13
In, the speaker standard pattern of the other party (caller) stored in the buffer is registered with the registrant number (caller number) added to the registrant speaker putter area in the speaker pattern memory 9.

Thus, the word standard pattern of the user name and the registrant name and the speaker standard pattern of the user and the registrant are registered, and the various registration processes are completed. In this state, if the user wants to send a message to the specific registrant while he is away, the following OGM recording process is performed. FIG. 5 is a flowchart of an OGM recording processing operation performed by the word recognition unit 1, the voice analysis / synthesis unit 3, and the speaker recognition unit 8 under the control of the control unit 7. Hereinafter, the OGM recording processing operation will be described in detail with reference to FIG.

In step S21, it is determined whether the "record key" in the key input unit 10 has been pressed. As a result, if it is pressed, the process proceeds to step S22 and the OGM
The recording processing operation is started. In step S22, the voice analysis / synthesis unit 3 generates the synthesized voice guidance “Please say the name of the message partner”, and the receiver 12
Output from In step S23, the above-described "registrant name recognition mode" routine is executed by the word recognition unit 1. As a result, a registrant number (caller number) of the word standard pattern, which is a recognition result of the registrant name uttered by the user according to the synthetic voice guidance, is obtained.
At that time, when the matching distance obtained at the time of the above-described matching by the word recognition unit 1 is equal to or more than a threshold (including a case where a registrant name other than the registered registrant name is uttered), or If the mute state continues for a predetermined time or more, the registrant number is set to "0".

In step S24, the speech analysis / synthesis unit 3
Synthesized voice guidance "Please say a message"
Is generated and output from the receiver 12. Step S
At 25, the above-mentioned “OGM” is
As a result, the OGM voice data based on the OGM uttered by the user according to the synthetic voice guidance is registered in the OGM area in the recording / playback memory 6. In step S26,
In parallel with the step S25, the above-mentioned "user recognition mode" routine is executed by the speaker recognition section 8, and the OGM recording process is completed. After that, the above-mentioned standby state is entered. As a result, in step S25, the person who uttered the OGM is recognized as a user, and the user number (called number) of the speaker standard pattern as the recognition result is obtained. Then, the obtained user number (called number) and the registrant number (calling number) obtained in step S23 are added to the OGM voice data registered in step S25. .

In this way, the OGM transmitted when a specific registrant receives an incoming call to the answering machine during absence.
Is recorded, the user presses the "answering machine key" on the key input unit 10 to put the answering machine in the answering machine standby state and go out. Then, when there is an incoming call to the answering machine in the answering machine standby state, the next incoming call processing is performed. FIG. 6 is a flowchart of the incoming call processing operation performed by the word recognition unit 1, the voice analysis / synthesis unit 3, and the speaker recognition unit 8 under the control of the control unit 7. Hereinafter, the incoming call processing operation will be described in detail with reference to FIG.

At step S31, it is determined whether or not the telephone line 14 has been connected to the answering machine. As a result, if connected, the process proceeds to step S32. In step S32,
The voice analysis / synthesis unit 3 generates synthesized voice guidance “Which of XXX, YYY, ZZZ to call” and sends it to the line connection unit 13. At this time, the audio signals relating to the user names “XXX”, “YYY” and “ZZZ” are output from the user name area in the echo back sound memory 4 based on the link information stored in the internal memory. It is generated based on the read-back echo-back sound audio data of the user name. On the other hand, the voice signal of the sentence “Who should you call?” Following the user name is generated based on the voice synthesis parameters stored in the synthesis memory 5.

In step S33, the above-described "user name recognition mode" routine is executed by the word recognition unit 1. As a result, a user number (called number) of the word standard pattern, which is a recognition result of the called name of the called party uttered by the calling party according to the synthetic voice guidance, is obtained. At that time, when the matching distance obtained in the above-described matching by the word recognition unit 1 is equal to or more than a threshold (including a case where a user name other than the registered user name is uttered), or for a predetermined time. When the mute continues, the user number is set to “0”. In step S34, in parallel with step S33, the speaker recognition unit 8
The above-mentioned "registrant recognition mode" routine is executed.
As a result, in step S33, the person who uttered the called party name is recognized as the registrant, and the registrant number (caller number) of the speaker standard pattern as the recognition result is obtained. At this time, if the minimum matching distance obtained at the time of the above-described matching by the speaker recognition unit 8 is equal to or larger than the threshold,
Set the registrant number to “0”.

In step S35, it is determined whether or not the "answering key" in the key input unit 10 has been pressed. As a result, if it is pressed, the process proceeds to step S36 to start the incoming call processing operation when the answering machine is on standby. On the other hand, if it is not pressed, the process proceeds to step S41, and the normal incoming call processing operation is started. Since the incoming call processing operation in this example is the incoming call processing operation in the answering machine standby state as described above, the process branches to step S36 in this step. In step S36, it is determined whether or not the OGM audio data to which the user number obtained in step S33 and the registrant number obtained in step S34 are added is stored in the OGM area in the recording / playback memory 6. Is checked. As a result, if there is the corresponding OGM audio data, the process proceeds to step S37,
If there is no M audio data, the process proceeds to step S38.

In step S37, the speech analysis / synthesis unit 3
As a result, the corresponding OGM audio data is read out from the recording / playback memory 6 and played back, and the OGM "I'm currently in a hotel PPP, so call me. The phone number is" 123-4567 ". , Please speak if you have a request "is generated and sent to the line connection unit 13. In step S38, since the corresponding OGM voice data does not exist, the voice analysis / synthesis unit 3 generates the normal synthesized voice guidance "Please answer your request because you are currently away". It is sent to the unit 13.

In step S39, the speech analysis / synthesis unit 3
Thus, the above-mentioned "ICM recording mode" routine is executed. As a result, the ICM voice data based on the ICM uttered by the caller in accordance with the OGM or the synthetic voice guidance is added to the step S as described above.
The caller's registrant number obtained in step 34 and the called party's user number obtained in step S33 are added and registered in the ICM area in the recording / playback memory 6.

In step S40, the above-mentioned "registrant pattern analysis mode" routine is executed by the speaker recognizing unit 8 in parallel with step S39, and the incoming call processing operation ends. After that, the above-mentioned standby state is entered. As a result, a speaker standard pattern is calculated from the voice signal of the caller transmitted via the telephone line 14, and is temporarily stored in the buffer. The calculation of the speaker standard pattern performed in this step is performed in preparation for a later registration as a registrant by a user's instruction because the caller is an unregistered person.

On the other hand, if it is a normal incoming call processing operation, the process branches to step S41 in step S35. In step S41, the voice analysis / synthesis unit 3 uses the user name (for example, “YYY”) associated with the user number obtained in step S33 and the registrant name (eg, “YYY”) obtained in step S34. For example, the synthesized voice guidance “AA” into which “AAA”) is inserted.
It is a call from A to YYY. "
2 and terminates the incoming call processing operation. afterwards,
The above-mentioned standby state is entered.

Note that the registrant name "AA"
The audio signal relating to "A" and the user name "YYY" is generated based on the echo back sound audio data of the registrant name or the user name read from the registrant name area or the user name area in the echo back sound memory 4. Because

Thereafter, when a call is started by the user "YYY", the start of the call is determined in step S12 in FIG. 4, and the above-described call registration processing operation is started.

When the user who returns home plays the ICM transmitted to himself while he is away, the following ICM playback processing is performed. FIG. 7 is a flowchart of an ICM reproduction processing operation performed by the word recognition unit 1, the voice recognition / synthesis unit 3, and the speaker recognition unit 8 under the control of the control unit 7. Hereinafter, the ICM reproduction processing operation will be described in detail with reference to FIG.

In step S51, it is determined whether the "play key" in the key input section 10 has been pressed. As a result, if it is pressed, the process proceeds to step S52 and the ICM
The reproduction processing operation is started. In step S 52, the speech analysis / synthesis unit 3 generates the synthesized speech guidance “Please say something” and outputs it from the receiver 12. In step S53, the above-described "user recognition mode" routine is executed by the speaker recognition unit 8. As a result, a user number (called number) of a speaker standard pattern, which is a recognition result of an arbitrary sentence uttered by the user according to the synthetic voice guidance, is obtained and transmitted to the control unit 7. At this time, if the minimum matching distance obtained in the above-described matching by the speaker recognition unit 8 is equal to or larger than the threshold, the user number is set to “0”.

In step S54, the recording / playback memory 6
Are searched, and the number of ICMs to which the user number (called number) obtained in step S53 is added is counted for each added registrant number (calling number). Is done. In step S55, based on the count result obtained in step S54 by the voice analysis / synthesis unit 3, a caller name (for example, "BBB", "CCC") related to the registrant number (caller number). ",
“Other (registrant number is“ 0 ”)” and the number of ICMs to which the registrant number (caller number) is added (for example, “2”,
“1”, “2”) and the synthesized voice guidance “BB
Two cases from B, one case from CCC, and two other cases ”are generated and output from the receiver 12.

In step S 56, synthesized voice guidance “Please say the name of the person you want to hear” is generated and output from the receiver 12. In step S57, the above-described "registrant name recognition mode" routine is executed by the word recognition unit 1. As a result, a registrant number (caller number) of the word standard pattern, which is a recognition result of the caller name (registrant name) of the other party uttered by the user according to the synthetic voice guidance, is obtained and transmitted to the control unit 7. You. At this time, if a registrant name that has not been registered is uttered, or if the mute state continues for a predetermined time or more, the registrant number (caller number) is set to “0”.

At step S58, the I number to which the user number (called number) obtained at step S53 is added.
From the CMs, the ICM to which the registrant number (caller number) obtained in step S57 is added is selected. In step S59, the above-described "ICM reproduction mode" routine is executed by the voice analysis / synthesis unit 3. As a result, the I selected in step S58
The CM (ICM from the specific registrant desired by the user to the user) is output from the receiver 12 by a synthesized voice.

In step S60, it is determined whether the "registration key" in the key input section 10 has been pressed. As a result, if pressed, the process proceeds to step S61. If not pressed, the process proceeds to step S64. In step S61, the voice analysis / synthesis unit 3 generates the synthesized voice guidance "Please say the name of the other party", and the receiver 1
2 output. In step S62, the word recognition unit 1
Thus, the above-mentioned “registrant name registration mode” routine is executed. As a result, the word standard pattern of the registrant name uttered by the user according to the synthetic voice guidance is registered with the registrant number (caller number) added to the registrant name pattern area in the word pattern memory 2.

In step S63, the above-mentioned "registrant pattern registration mode" routine is executed by the speaker recognition section 8. As a result, the speaker standard pattern of the new registrant stored in the buffer at step S40 in the flowchart of the incoming call processing operation is read, the registrant number is added, and the registrant in the speaker pattern memory 9 is read. Registered in the speaker pattern area. The registration of the new registrant name and registrant pattern carried out in the above-mentioned steps S61 to S63 is carried out by calling the IC from an unspecified partner without uttering a specific name in step S57.
M (ICM with registrant number (caller number) “0” added)
Is reproduced when the information about the other party is newly registered.

In step S64, the user determines whether or not there is any ICM to be reproduced. As a result, if there is, the process returns to step S56 to start ICM reproduction of the next registrant. On the other hand, if not, the ICM reproduction processing operation is terminated. After that, the above-mentioned standby state is entered.

Thus, a series of “initial registration processing”, “call registration processing”, “OGM recording processing”, “incoming processing”, and “I
By executing the “CM playback process”, the ICM from the specific registrant or the non-registrant desired by the user is selected from among the ICMs from a plurality of callers received while the user is away.
Only you can select and play.

As described above, in this embodiment, the word recognition unit 1 registers / recognizes a user name or a registrant name.
And a speaker recognition unit 8 for registering / recognizing a user or a registrant. Further, based on the recognition result by the word recognition unit 1 or the speaker recognition unit 8, the registrant number (caller number)
It has a voice analysis / synthesis unit 3 for recording / reproducing OGM or ICM to which a pair of an OGM or an ICM and a user number (called number) is added.

When there is an incoming call in the answering machine standby state, the control unit 7 automatically answers and obtains a user number (called number) by recognizing the user name by the word recognizing unit 1.
A registrant number (caller number) is obtained by speaker recognition by the speaker recognition unit 8. Then, an OG to which a pair of the obtained user number (called number) and registrant number (calling number) is added.
By reading M from the recording / playback memory 6 and transmitting it to the caller, OG to the specific registrant desired by the user is performed.
Transmit M. On the other hand, the ICM transmitted from the calling party to the called party in response to the OGM is recorded with a pair of a registrant number (calling party number) and a user number (called party number).

When playing back the ICM recorded in this way, the user number (called number) of the specific user who performs the ICM reproduction is obtained by the speaker recognition by the speaker recognition unit 8, while the word recognition unit 1 obtains the user number. Registrant number of ICM sender
(Caller ID). Then, the obtained ICM to which the pair of the registrant number (caller number) and the user number (called number) is added is reproduced by the voice analysis / synthesis unit 3 and output to the siever 12 by The ICM from the specific registrant desired by the specific user is reproduced.

Therefore, even when a plurality of users use the answering machine, when there is an incoming call in the answering machine standby state, the caller and the called party are automatically determined, and the determination result is obtained. Is transmitted to the calling party, while the ICM corresponding to the determination result can be presented to the called party. That is, according to this embodiment, in message transmission / reception between a plurality of users and a plurality of registrants, transmission of OGM from a specific user to a specific registrant and presentation of ICM from the specific registrant to the specific user are enabled. It is.

When registering a registrant pattern at that time, the registrant must use the “registrant pattern registration mode”.
Sometimes it is only necessary to utter a call or ICM sentence, and it is not necessary to take the troublesome procedure of uttering a keyword in accordance with a user's instruction, so that it can be easily implemented.

The algorithms of the initial registration processing operation, the call registration processing operation, the OGM recording processing operation, the incoming call processing operation, and the ICM reproduction processing operation in the present invention are not limited to the algorithms in the above embodiment.

In the incoming call processing operation in the above embodiment, the OGM is simply transmitted in step S37, and
The location and telephone number of the called party are transmitted to the calling party. However, the present invention is not limited to this, and the call may be forwarded automatically after transmitting the location of the called party by OGM.

In the above embodiment, the user number and the registrant number may be numbers indicating a specific user or a registrant, respectively, as described above, or may be stored in the word pattern memory 2 or the speaker pattern memory 9. A number indicating the order of storage may be used. However, when the number indicating the storage order is used as the user number or the registrant number, the number of each word standard pattern stored in the word pattern memory 2 and each speaker standard number stored in the speaker pattern memory 9 are used. It is necessary to store the association with the pattern number, for example, in the internal memory of the control unit 7.

[0104]

As is clear from the above description, in the answering machine of the first invention, during recording, the control unit controls the word recognition unit, the speaker recognition unit, and the voice analysis / synthesis unit so that the voice of the specific user is recorded. OG of OGM from a specific user to a specific registrant obtained by the voice analysis / synthesis unit based on
The user number and the registrant number obtained by the word recognition unit and the speaker recognition unit are added to the M voice data and stored in the recording / playback memory. While registering the word standard pattern of the user name and registrant name generated by the above in the word pattern memory, the speaker standard pattern of the user and the registrant generated by the analysis of the speaker recognition unit is registered in the speaker pattern memory. With this arrangement, based on the voice of the specific user, the OGM assigned from this specific user to the specific registrant can be recorded with the sender / receiver information added.

Therefore, according to the present invention, when there are a plurality of users who want to record OGMs or when they want to record different OGMs for each receiver, a combination of a specific user who is a sender and a specific registrant who is a receiver OGM can be recorded separately for each.

Also, in the answering machine according to the second invention, when there is an incoming call in the answering machine standby state, the control section automatically answers and controls the word recognition section, the speaker recognition section and the voice analysis / synthesis section, Based on the voice of the caller (specific registrant) from the telephone line, the user number of the called party and the registrant number of the caller obtained by the word recognition unit and the speaker recognition unit are added. The audio data for OGM is read out from the recording / playback memory, reproduced as an audio signal, and sent out to the telephone line. Therefore, the audio analysis / synthesis unit previously transmits a specific user to a specific partner (specific registrant). If the OGM is stored in the recording / playback memory with transmission / reception information added thereto, the specific user (specific called party) is determined based on the user number and the registrant number obtained by the voice of the specific registrant. ) To specific registrant (specific caller) It can send the OGM automatically.

At this time, when registering / recognizing the speaker standard pattern of the specific registrant, the registration / recognition is automatically performed by the call voice uttered by the specific registrant. There is no need to prompt specific registrants to speak a keyword each time a pattern is registered / recognized.

That is, according to the present invention, the specific user
The OGM from the (specified called party) to the specified partner (specified calling party) can be easily and automatically transmitted through the telephone line.

In the answering machine of the third invention, when there is an incoming call from the telephone line in the answering machine standby state, the control section controls the word recognition section, the speaker recognition section and the voice analysis / synthesis section. The word recognition unit and the speaker are added to the ICM voice data of the ICM from the caller to the called party obtained by the voice analysis / synthesis unit based on the voice of the caller (specific registrant). Since the registrant number of the calling party and the user number of the called party obtained by the recognition unit are added and stored in the recording / playback memory,
While the word standard patterns of the user name and the registrant name generated by the analysis of the word recognition unit are registered in the word pattern memory in advance, the speaker standard patterns of the user and the registrant generated by the analysis of the speaker recognition unit are spoken. If the ICM assigned to the specific user is transmitted from the specific registrant based on the voice of the specific registrant,
You can add recipient information and record.

Therefore, according to the present invention, even when there are a plurality of callers (registrants) or callees (users) of the received ICM, the specific registrant who is the caller and the receiver are different. ICM can be recorded separately for each group with a specific user.

Also, in the answering machine of the fourth invention, at the time of reproduction after the end of the answering machine standby, the control section controls the word recognition section, the speaker recognition section and the voice analysis / synthesis section,
Based on the voice of the specific user, the ICM voice data to which the registrant number of the calling party and the user number of the called party obtained by the word recognition section and the speaker recognition section are added is subjected to the voice analysis / Since the synthesizing unit reads out from the recording / playback memory and reproduces it as an audio signal and sends it to the receiver, a specific partner (specific registrant) input from the telephone line in advance when receiving an incoming call in the answering machine standby state is transmitted to the specific user. If the ICM is added to the transmission / reception information by the voice analysis / synthesis unit and stored in the recording / playback memory, the ICM is added to the user number and the registrant number of the specific user obtained by the voice of the specific user. Based on the specific registrant
The ICM from the (specific caller) to the specific user (specific callee) can be automatically reproduced.

At this time, when registering / recognizing the speaker standard pattern of the specific registrant, the registration / recognition is automatically performed by the call voice uttered by the specific registrant. There is no need to prompt specific registrants to speak a keyword each time a pattern is registered / recognized.

That is, according to the present invention, the specific partner
The ICM from the (specific caller) to the specific user (specific callee) can be easily and automatically transmitted based on the utterance of the registrant number.

In the answering machine of the fifth invention, when an incoming call is received in the answering machine standby state, the control unit:
The word recognition unit, the speaker recognition unit, and the voice analysis / synthesis unit are controlled to automatically transmit OGM from a specific user to a specific partner via the telephone line.
A registrant number and a user number obtained by the word recognition unit and the speaker recognition unit are added to ICM voice data of the ICM transmitted from the caller to the called party via the telephone line by the combining unit. Is stored in the recording / playback memory, and at the time of playback after the answering machine standby is over, the registrant number and the user number obtained based on the voice of the specific user by the word recognition unit and the speaker recognition unit are added. Since the read ICM is read out and reproduced as an audio signal and output from the receiver, the ICM from the specific registrant (specific caller) to the specific user (specific callee) is automatically reproduced. Can output audio. Therefore, according to the present invention, the OGM from a specific user to a specific partner and the ICM from a specific partner to a specific user can be easily and automatically recorded / reproduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of an answering machine of the present invention.

FIG. 2 is a flowchart of an operation mode when various processes are performed by the answering machine shown in FIG. 1;

FIG. 3 is a flowchart of an initial registration processing operation.

FIG. 4 is a flowchart of a call registration processing operation.

FIG. 5 is a flowchart of an OGM recording processing operation.

FIG. 6 is a flowchart of an incoming call processing operation.

FIG. 7 is a flowchart of an ICM reproduction processing operation.

FIG. 8 is a block diagram of a conventional answering machine.

[Explanation of symbols]

1. Word recognition unit 2. Word pattern memory 3. Speech analysis / synthesis unit 4.
Echo back sound memory, 6 ... Recording and playback memory,
7: control unit, 8: speaker recognition unit,
9: Speaker pattern memory, 10: Key input unit,
13 ... line connection part, 14 ... telephone line.

Continuation of the front page (56) References JP-A-57-131193 (JP, A) JP-A-61-39756 (JP, A) JP-A-61-43862 (JP, A) JP-A-61-109356 (JP, A) , A) JP-A-61-186053 (JP, A) JP-A-63-67949 (JP, A) JP-A-64-86742 (JP, A) JP-A-3-167947 (JP, A) 5-300215 (JP, A) Japanese Utility Model 62-14881 (JP, U) (58) Fields investigated (Int. Cl. ⁶ , DB name) G10L 3/00-3/02 H04M 1/64-1 / 65 H04M 3/42

Claims (5)

(57) [Claims] At the time of registration, the voice of the input name of a user or a registrant is analyzed to obtain a word standard pattern of a user name or a registrant name, and a user number or a registrant number is added and registered in a word pattern memory. While
When recognizing, a word recognizing unit for obtaining a user number or a registrant number added to a word standard pattern of a user name or a registrant name obtained by recognizing the voice of the input name of the user or the registrant, The voice of the input user or registrant is analyzed to obtain a speaker standard pattern, and the above user number or registrant number is added and registered in the speaker pattern memory. Speaker recognition unit for finding the user number or registrant number added to the speaker standard pattern obtained by recognizing the voice generated by the speaker, and the voice of the outgoing message from the specified user to the specified registrant The outgoing message voice data obtained by analyzing and compressing the A speech analysis / synthesis unit to be stored in memory, during recording, the word recognition unit, and controls the speaker recognition unit and speech analysis / synthesis unit, based on the voice of a particular user,
An answering machine having a control unit for adding and recording a user number and a registrant number corresponding to outgoing message voice data of the outgoing message from the specific user to the specific registrant. 2. The answering machine according to claim 1, wherein the voice analysis / synthesis unit records and reproduces specific outgoing message voice data based on the input user number and registrant number during reproduction. The control unit automatically reads out from the telephone line in the answering machine standby state when the call is received from the telephone line, and the word recognition unit, the speaker recognition unit, and the voice analysis / The control unit controls the synthesizing unit to read outgoing message voice data to which the user number and the registrant number obtained based on the voice of the caller from the telephone line are added, and reproduce the voice signal to reproduce the voice signal. An answering machine characterized by being sent to a telephone. 3. At the time of registration, the voice of the input name of the user or registrant is analyzed to obtain a word standard pattern of the user name or registrant name, and a user number or registrant number is added and registered in the word pattern memory. While
When recognizing, a word recognizing unit for obtaining a user number or a registrant number added to a word standard pattern of a user name or a registrant name obtained by recognizing the voice of the input name of the user or the registrant, The voice of the input user or registrant is analyzed to obtain a speaker standard pattern, and the above user number or registrant number is added and registered in the speaker pattern memory. Speaker recognition unit for obtaining the user number or registrant number added to the speaker standard pattern obtained by recognizing the voice generated by the speaker, and the voice of the message transmitted from the specific registrant to the specific user The voice data for a message message obtained by analyzing and compressing the message is added to the corresponding registrant number and user number for recording and playback. A voice analysis / synthesis unit to be stored in the memory, and automatically answers when there is an incoming call from the telephone line in the answering machine standby state, and controls the word recognition unit, the speaker recognition unit, and the voice analysis / synthesis unit, and Based on the input caller's voice, the caller's registrant number and the callee's user number are added to the voice message message data of the message from the caller to the callee. An answering machine, characterized by comprising a control unit for making a recording. 4. The answering machine according to claim 3, wherein the voice analyzing / synthesizing unit records and reproduces specific voice message voice data based on the input registrant number and user number during reproduction. The control unit controls the word recognition unit, the speaker recognition unit, and the voice analysis / synthesis unit to perform the identification at the time of the reproduction after the end of the answering machine standby. An answering machine characterized by reading out voice data for a specific message to which a registrant number and a user number obtained based on a user's voice are added, reproducing the voice data, and transmitting the voice signal to a receiver. . 5. The voice mail according to claim 2, wherein the voice analysis / synthesis unit analyzes and compresses the voice of the message transmitted from the specific registrant to the specific user. The data is stored in the recording / playback memory with the corresponding registrant number and user number added thereto, and at the time of playback after the answering machine standby is over, a specific message message is generated based on the entered registrant number and user number. The voice data is read out from the recording / playback memory and reproduced as a voice signal. When the control unit automatically answers in the answering machine standby state, based on the input caller's voice, The voice message data of the message from the caller to the called party input in response to the outgoing message sent to the telephone line; The user's registrant number and the called party's user number are added and recorded, and at the time of playback after the answering machine standby is over, the registrant number and user number obtained based on the voice of the specific user are added. An answering machine characterized in that voice data for a specific message is read, reproduced as a voice signal, and transmitted to a receiver.