JP2005175649A - Voice notice apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice notice apparatus for carrying out a notice in a voice with a property in response to the attribute of a called party. <P>SOLUTION: The voice notice apparatus 2 is provided with: a storage means 22 for storing notice data 221 including a telephone number and a customer attribute; a speech contents control means 23 for determining the property of voice in the case of synthesizing a sentence of call contents into the voice on the basis of the customer attribute of the notice data 221; a voice synthesis means 24 for synthesizing the sentence of the call contents with the voice having the property determined by the call contents control means 23; and a voice transmission means 25 for dialing the telephone number of the notice data 221 and transmitting the voice synthesized by the voice synthesis means 24. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a voice notification device, and more particularly, to a voice notification device that makes a voice notification based on a voice having a property corresponding to an attribute of a call partner or a content of notification.

In financial institutions, etc., a system for automatically making a dunning call is adopted. For example, a system is known in which dunning data is converted into voice, automatically called to a dunning partner, and automatically notified or dunned to the dunning partner. (For example, refer to Patent Document 1). In addition, a system is known in which dunning list change information is transferred to a dunning agent by telephone in real time, and the dunning list is updated in real time and reflected in the dunning service of the agent. (For example, refer to Patent Document 2).
JP-A-4-354250 JP 2002-7692 A

  In the above-described system for automatically calling a dunning call, a dunning person can be dunned before a predetermined time limit. However, only predetermined voice data is repeatedly transmitted regardless of the characteristics (attributes) of the dunning target person such as age, gender, and personality, and dunning according to the attributes of the dunning target person cannot be performed. For example, if the reminder is an elderly person, it is preferable to repeat the sound multiple times with a large and slow sound. Moreover, if the reminder is a young person, a slightly quicker and simpler sentence is preferable. For this reason, there is a possibility that the contents cannot be clearly communicated to the reminder, and the discomfort such as long (or short) call time, too fast (or too late) call speed, or too long content. You may not be able to dunn without giving

  An object of the present invention is to provide a voice notification device that performs notification by voice having a property corresponding to the attribute of the other party of the call.

  Another object of the present invention is to provide a voice notification device that performs voice notification according to the content of notification according to the attribute of the other party of the call.

  The voice notification device according to the present invention includes a storage means for storing notification data including a telephone number and a customer attribute, and a voice message when synthesizing a sentence that is a call content based on the customer attribute of the notification data. Call content control means for determining the nature, voice synthesis means for synthesizing the text that is the content of the call to speech of the nature determined by the call content control means, and calling the telephone number of the notification data, Voice transmitting means for transmitting the voice synthesized by the voice synthesizing means.

  Further, the voice notification device of the present invention determines the configuration of the text that is the content of the call based on the storage means for storing the notification data including the telephone number and the customer attribute, and the customer attribute of the notification data. Based on the call content control means for creating the text as the call content based on the above, the speech synthesis means for synthesizing the text as the call content created by the call content control means, and the telephone number of the notification data Voice transmitting means for calling and transmitting the voice synthesized by the voice synthesizing means.

  According to the voice notification device of the present invention, the content of the dunning or the like is notified by a voice having properties according to the characteristics (attributes) such as the age, sex, unmarried / married, etc. be able to. For example, if the other party of the call is an elderly man, the content can be notified by the voice of a young woman, the volume is increased, and the voice is slow. If the other party of the call is a young woman, the content can be notified with a voice of a young man, a normal volume, and a somewhat quick voice. As a result, the content can be clearly communicated to the other party of the call without causing discomfort.

  Further, according to the voice notification device of the present invention, it is possible to make a voice notification based on the content of the notification according to the attributes of the other party of the call, such as age, sex, unmarried / married. For example, if the other party of the call is an elderly person, the contents can be notified by repeating the polite sentence about three times. If the other party of the call is a young person, the contents can be notified by repeating the concise sentence about twice. As a result, the content can be clearly communicated to the other party of the call without causing discomfort.

  FIG. 1 is a block diagram of a voice notification system and shows a configuration of a voice notification system including a voice notification device of the present invention. The voice notification system includes a data management device 1, a voice notification device 2, a telephone 3 owned by a person to be notified or a customer, and networks 41 and 42 connecting them. The data management device 1 includes a computer provided in a financial institution such as a bank. The voice notification device 2 includes a computer provided in the financial institution or a notification agent. The telephone 3 is a subscriber telephone such as a fixed telephone or a mobile phone. The network 41 is composed of the Internet (or public line), for example, and the network 42 is composed of the public line (or Internet), for example. The networks 41 and 42 may be the same.

  The data management apparatus 1 includes a notification data transmission unit 11 and a notification data database (DB) 12. The notification data DB 12 stores a plurality of notification data 221 (see FIG. 2A) created in advance. For example, the notification data 221 is created in advance for each lending for each lending based on a lending list (not shown) during execution of daily work. The notification data transmission unit 11 reads one notification data 221 from the notification data DB 12 in a predetermined order, and for example, notification data for customers who have reached one week before the deadline (excluding customers who have already paid). 221 are read in the order of arrival (in this case, the attribute “type” is notification), or the notification data 221 for customers who have not paid for the past due for one week has passed (in this case, the attribute “Type” is a reminder) and requests the voice notification device 2 to notify the customer by voice based on the notification data 221 via the network 41. As will be described later, the notification data 221 includes, for example, the telephone number of the telephone 3 owned by the customer and customer attributes. Note that the notification data 221 is not limited to notifications and reminders, and may be created based on a customer list, for example, to notify advertising data or to sell special sale data. In this case, the attribute “type” is notification, and the charge amount and penalty are “empty”.

The voice notification device 2 receives the notification data 221 from the data management device 1 and sends a voice notification to the customer (phone 3 owned by the customer) via the network 42 based on the notification data 221. Do. For this purpose, as shown in FIG. 1, the voice notification device 2 includes a notification data receiving unit 21, a storage unit 22, a call content control unit 23, a voice synthesis unit 24, and a voice transmission unit 25. The notification data receiving unit 21 includes a known communication unit that communicates with the data management apparatus 1 via the network 41 to obtain the notification data 221. The voice transmission means 25 communicates with the customer's telephone 3 via the network 42, and uses a known voice communication means (telephone communication means) for notifying the customer (phone 3 owned by the customer) by voice. Prepare.

  The notification data receiving means 21 receives notification data 221 including a telephone number and customer attributes together with a notification request from the data management apparatus 1 and stores it in the storage means 22. That is, the one piece of notification data 221 is stored at the end of the notification data storage area (not shown) of the storage means 22 in the order of reception.

  The storage means 22 stores notification data 221 including a telephone number and customer attributes. An example of the notification data 221 is shown in FIG. One row (one record) in FIG. 2A is one piece of notification data 221. In the notification data storage area, one or a plurality of notification data 221 is stored in the order of reception, for example. The one piece of notification data 221 at the head of the area is the processing target of the voice call. If a connection cannot be established even after a predetermined time has elapsed since the call was made, the notification data 221 is taken out from the beginning of the area and stored again at the end.

  In this example, the customer attributes of the notification data 221 that are mainly considered (used) are age, sex, marriage (unmarried / married), or a combination thereof. The attributes used are not limited to these, and occupations may be used. One piece of notification data 221 stores, for each notification ID, the customer ID, telephone number, age, sex, marriage (unmarried / married), occupation, type, deadline, billing amount, and penalty. The notification ID and customer ID are unique in the system. The type is “notification” if the current time point (or the time when the voice call is likely to be made) is before the deadline, and “reminder” otherwise. The penalty is, for example, the rank information of the customer as viewed from a financial institution, for example, “0” means that all payments were made before the deadline in the past, “1” means that payments were past due, but “ “2” indicates a worse state of “1”, and “3” indicates past due and currently unpaid. A customer with a penalty of “0” is a good customer and may be excluded from the notification target. For example, a customer with a penalty of “1” may be notified one week before, and a customer with a penalty of “3” may be notified one week and three days ago.

  In addition, the storage unit 22 stores a call history 222 including call state information indicating an end state of the call or a call disconnection position for each call. The call history 222 is a call status for all calls made in the past week or month, for example. The period for storing the call history 222 is, for example, a period sufficiently longer than the interval from the previous notification (or dunning) to the next notification (or dunning) in the financial institution (or the voice call system). .

  An example of the call history 222 is shown in FIG. One call history 222 stores the date and time and the call status for each call. The call status includes a notification ID, a customer ID, and call status information. From the notification ID and the customer ID, it is possible to determine which case of which customer. The call state information indicates a call end state or a call disconnect position. That is, the last symbol “ZZ” in the call status information indicates that the call has been made to the end of the sentence, and “2B” indicates that the call is made by the customer during the period in which the sentence B constituting the second repeated sentence is flowing. Indicates disconnected.

  The call content control unit 23 reads one piece of notification data 221 from the storage unit 22 (the head thereof), and synthesizes a sentence that is the content of the call into speech based on the customer attribute of the notification data 221. The nature of the speech is determined and sent to speech synthesis means 24. The nature of the voice is the voice quality, speed, volume, or a combination thereof. The nature of speech is not limited to these.

Further, the call content control means 23 determines the composition of the text as the call content based on the customer attribute of the notification data 221 as well as the nature of the voice, and creates the text as the call content based on this. To the speech synthesis means 24. The composition of the sentence is the word order of the sentence, the length of the sentence, the number of repetitions, or a combination thereof. The composition of sentences is not limited to these.

  The call content control means 23 determines the nature of the voice and determines the structure of the text that is the call content. For this purpose, the call content control means 23 includes a parameter table 231, a target person type table 232, and a syntax table 233. An example of these tables 231 to 233 is shown in FIGS. That is, FIG. 3 shows the parameter table 231, FIG. 4A shows the target person type table 232, and FIG. 4B shows the syntax table 233. The contents of these tables 231 to 233 are determined in advance, but may be changed as appropriate.

  The parameter table 231 defines the voice parameters for each target person type. The voice parameters in this example include voice quality, speed, volume, notification text, dunning texts 1 to 3 and (dunning) repetition count. The notification sentence defines the structure of the sentence that is the content of the call when the attribute “type” of the notification data 221 is “notification”. The dunning text defines the composition of a text that is the content of a call when the attribute “type” of the notification data 221 is “Dunning”. The number of repetitions determines the number of repetitions of the dunning sentence when the attribute “type” of the notification data 221 is “reminding” (the sentence may be different for each repetition). The target person type table 232 defines attributes of customers belonging to the target type for each target person type. In this example, the target person types are classified into six types “1 to 6” shown in FIG. 4A including the answering machine. The target person type is not limited to these, and occupations may be considered. The syntax table 233 stores sentence candidates (sentences) together with their symbols (sentence symbols). Sentence candidates (sentences) are not limited to this.

  The call content control means 23 first determines the type of the target person by referring to the target person type table 232 using the customer attribute of the notification data 221. For example, if the customer's age is “60 years old or older”, as shown in FIG. 4A, the type is “elderly” regardless of gender and unmarried / married. Next, the call content control means 23 determines the parameter by referring to the parameter table 231 using this type. For example, if the customer type is “elderly”, as shown in FIG. 3, the voice quality is “young woman”, the speed is “slow”, the volume is “large”, the syntax of the notification sentence is “A + B + C + E + F + G”, The syntax of the dunning sentences 1 to 3 is “A + B + D + E + F + G”, and the number of repetitions of the sentence is “3 times”. Furthermore, the call content control unit 23 refers to the syntax table 233 using this parameter and determines the sentence. For example, in the case of this notification sentence, sentences of A, B, C, E, F, and G are read in this order from the syntax table 233 and connected to create a sentence that is the content of the call. The difference between the notice text and the dunning text is that C “·· is due for payment” in the notice text is replaced with D “·· is due for payment” in the reminder text. It is in.

  In this example, as shown in FIG. 4B, the sentence elements stored in the syntax table 233 are A for “company name”, B for “target property”, C for “notification”, and D for “Dunning”, E is “billing amount”, F is “penalty”, and G is “knot”. That is, it consists of the minimum necessary elements. Therefore, even if the notification sentence and the dunning sentence have the longest configuration, the notification sentence and the reminder sentence can be a simple sentence composed of the minimum necessary information.

  In this example, the voice quality is the voice of a young woman for any customer type (excluding “answer machine”) to facilitate acceptance of reminders and the like. The speed is generally “slow” when the customer type is “elderly” and the hearing loss generally declines, and it is not frustrated when “young (male)” and “young (female)”. “Slightly faster” (also “answering machine”). The volume is set to “high” in the case of “elderly” for the same reason as described above.

In this example, the syntax of the notification sentence (the sentence when the attribute “type” is notification) is “
A + B + C + E + F + G ”, which is common to all customer types. On the other hand, the syntax of the dunning text (text when the attribute “type” is dunning) is made different depending on the customer type (that is, customer attribute), and the number of repetitions is also different. In addition, the syntax (word order) of the sentence is changed for each repetition. Note that the syntax of the notification text may be variously changed similarly to the syntax of the dunning text.

  Specifically, if the customer type is “elderly”, there is a high possibility of being missed, so the syntax is the longest sentence, the number of repetitions is the largest, for example “3 times”, and the same sentence And the word order is the normal word order (“A + B + D + E + F + G”).

  If the customer type is “middle-aged”, it is better to give the conclusion after informing the conclusion first, so the first syntax is a little shorter and the number of repetitions is normal, for example “twice” Finally, the longest sentence is repeated, and the word order of the first dunning sentence (dunning sentence 1) is such that details or penalty (F) is omitted.

  If the customer type is “Housewife”, it is sensitive to interest rates, so it is better to tell the whole after informing this first, so the first syntax is the longest sentence, and the number of repetitions is normal eg “Twice” is set to repeat the longest sentence at the end, and the word order of the first dunning sentence is assumed to be placed in front of the interest rate (F) that would be of high interest as much as possible.

  In this example, in consideration of the attribute “occupation”, in the case of “˜60 years old”, “woman”, “married” and “unemployed”, the target person type is “housewife (so-called housewife) (See FIG. 4A). In this way, in determining the target person type, a specific attribute that is not considered in other target person types may be considered for one target person type. In this example, “unmarried / married” is also considered only in the target person type “housewife”.

  If the customer type is “Young (male)”, if the sentence is long, it will not be heard until the end, so the syntax will be the shortest sentence, the number of repetitions will be normal, for example “2 times”, the last The word order of the first reminder sentence is such that only the requirements (A, D, E, G) are arranged in that order.

  When the customer type is “young (female)”, even if the sentence is long, there is almost no customer who hears to the end and misses, so the syntax is the longest sentence and the number of repetitions is minimized. The word order of the first dunning sentence is the normal word order.

  In the sentence structure, the sentence structure with the second and subsequent repetitions is the longest (ie, detailed) sentence. This is because such sentences are for customers who have missed the first (and / or second) sentence, and the parts they missed are not uniform, and the awareness that such customers have missed This is because it is normal to listen to the end.

  Also, when connected to the telephone number of the notification data 221, if the other party is an answering machine, it will be the voice of a young woman so that it can be heard easily, and it can be rehearsed many times, so it is a little faster, at normal volume, The syntax is the longest sentence, and the number of repetitions in the sentence structure is “one”. Whether or not the other party is an answering machine is determined, for example, by detecting a sound that signals the start of a message (a signal sound called beep).

In this example, the call content control means 23 actually determines the nature of the voice based on the call state information for the telephone number in the call history 222 in addition to the customer attribute of the notification data 221. The composition of the sentence that is the contents of the call is determined, and the sentence that is the contents of the call is created based on this. That is, since an attribute that is a general classification does not always apply to each individual, the determined voice property and sentence structure are corrected (corrected) based on the call state information.

  For example, when the notification ID shown in FIG. 2 (A) is “A0300102” and the customer ID is “K120589”, the attributes are “45 years old”, “female”, and “married”. The person type is “housewife” from the person type table 232. Therefore, from the parameter table 231 of FIG. 3, “notification” of a sentence composed of “A + B + C + E + F + G” is performed as a parameter with a voice quality of “young man” at a speed of “normal” and a volume of “normal”. Is "A + D + F + B + E + G", and the second time is to "remind" a sentence consisting of "A + B + D + E + F + G".

  However, when the customer refers to the call history 222 in FIG. 2B, the record is “2B”. That is, it can be estimated that the customer understands the contents when he hears the first sentence and cuts when he finds that the second sentence is a repetition. Therefore, the repetition parameter for “housewife” is “twice”, but for this customer, the repetition parameter is corrected to “once”. Or the composition of the first reminder sentence for “housewife” is “A + D + F + B + E + G”, but for this customer, the composition is modified to “A + D + E + G”. As a result, the customer can be prompted without discomfort. If the customer record is “ZZ”, the repetitive parameters are not corrected.

  Conventionally, as shown in FIG. 6A, the audio parameter is fixed. For example, once the voice quality is set to “young woman”, this cannot be changed for each customer according to the customer attributes. For example, regarding the notification text, as shown in FIG. 6B, the text that is the content of the call was also fixed. For example, once it is set as “OO company ...”, it could not be changed for each customer in accordance with the customer attribute. The same applies to the dunning text. Actually, since it is difficult to use the notification sentence and the reminder sentence properly, the same sentence shown in FIG. 6B is often used.

  The voice synthesizing unit 24 receives the text that is the content of the call along with the nature of the voice from the call content control unit 23, and based on these, the text that is the content of the call created by the call content control unit 23 is converted into the call content control unit 23. Is synthesized into speech of the nature determined by. That is, a sentence composed of the designated word order, length, and number of repetitions is synthesized with the voice quality, speed, and volume of the designated voice.

  Here, as a means for speech synthesis, a well-known means for recording a voice of the owner of the voice to be used in advance and performing speech synthesis by a combination thereof can be used. That is, what is recorded by generating various sounds for a specific speaker is made into a database as (speech) synthesis basic data. Then, a synthesized speech based on the voice quality of a specific speaker is obtained by inputting a sentence to be reproduced into a program that performs speech synthesis processing using the synthesis basic data. As such means, for example, SpeeCAN-TTS of Arcadia Co., Ltd. and CHART of ATR Spoken Language Communication Research Institute are known. In this example, synthesis basic data (DB) of “young men” and “young women” created in advance by any means is provided in a memory (not shown) of the speech synthesis means 24.

  Further, as means for speech synthesis, a well-known means for performing speech synthesis at a frequency of voice quality to be reproduced can be used. That is, synthesized speech is obtained by reproducing speech (sentence to be reproduced) at a frequency suitable for the voice quality (male, female, age) desired to be reproduced. As such means, for example, FineSpeech of Animo Co., Ltd. is known.

The voice transmitting unit 25 calls the telephone number of the notification data 221 and, when connected, transmits the voice synthesized by the voice synthesizing unit 24 to the telephone 3 having the telephone number. Further, when connected, the voice transmitting unit 25 creates a call history 222 for the call, and at the end of the call history storage area (not shown) of the storage unit 22, the one call history 222 is stored. Accordingly, the call history 222 is created and stored for all calls. When the telephone number of the notification data 221 is called and not connected, the unconnected state may be created and stored as the call history 222.

  FIG. 5 shows a voice notification processing flow, and shows a voice notification process in the voice notification device 2 of the present invention.

  The notification data receiving means 21 receives the notification data 221 from the data management device 1 and stores it in the storage means 22 (notification data storage area) (step S1). When there is the notification data 221 in the storage means 22, the call content control means 23 reads it from the storage means 22 (step S2), obtains the target person type from the attribute of the customer (step S3), The parameter is obtained from the person type (step S4). Furthermore, the call content control means 23 refers to the call history 222 of the notification ID and customer ID, and corrects the parameter obtained in step S4 as necessary (step S5).

  Thereafter, based on the notification data 221 sent from the call content control means 23, the voice transmission means 25 calls the telephone number of the notification data 221 (step S6), and after being connected, It is detected whether or not the telephone 3 is an answering machine (state) (step S7). When the other party's telephone 3 is an answering machine, the voice transmission means 25 sends a notification to that effect to the call content control means 23. Based on this, the call content control means 23 corrects the parameter obtained in step S4 (step S8). If the other party's telephone 3 is not an answering machine, step S8 is omitted.

  The “voice quality” in the parameter obtained in step S4 or the parameter corrected in step S5 or S8 is sent from the call content control means 23 to the voice synthesis means 24. Based on this, the speech synthesis means 24 loads the corresponding synthesis basic data into its own memory (step S9). Further, the call content control means 23 sends the parameter (remaining part) to the speech synthesis means 24 (step S10). Based on this, the voice synthesizing means 24 loads the notification sentence or the dunning sentence (A etc.), which is the corresponding call content, into its own memory (step S11), and the sentence at the volume and speed according to the parameters. Speech synthesis is performed (step S12). As a result, the voice synthesized by the voice synthesizing means 24 is transmitted as voice to the telephone 3 of the other party connected by the voice transmitting means 25.

  After step S12, the speech synthesizer 24 checks whether the sentence is the last sentence of the notification sentence or the dunning sentence that is the content of the call (step S13). If it is not the last sentence, step S11 and subsequent steps are repeated. If it is the last sentence, the speech synthesizer 24 further checks whether the number of repetitions is the last number (step S14). If it is not the final number of times, the speech synthesizer 24 returns the object to be reproduced to the head sentence of the repeated sentence (step S15), and repeats step S11 and subsequent steps. If it is the last number, the process is terminated. Actually, the voice transmission unit 25 creates a call history 222 for the call before the processing is completed, and stores the call history 222 in a predetermined position in the storage unit 22.

From the above, for example, when the notification ID shown in FIG. 2A is “A0300101” and the customer ID is “K013456”, the attribute is “28 years old”, “male”, “single”. ), The target person type is “young (male)”. Therefore, “notification” of a sentence composed of “A + B + C + E + F + G” is made by synthetic voice on the telephone 3 from the parameter table 231 of FIG. “Reminder” of a sentence consisting of “A + D + E + G” for the second time and “A + B + D + E + F + G” for the second time is made by synthesized speech. The customer's call history indicates that he / she has heard to the end as shown in FIG. 2B, so the parameters are not modified.

  Further, when the notification ID shown in FIG. 2A is “A0300102” and the customer ID is “K120589”, the target person type is “housewife” from the attribute as described above, and the corresponding parameter is obtained. However, as described above, since the record in the call history 222 of this customer is “2B”, for example, for this customer, the repetition parameter is corrected to “once”, or the first dunning sentence Modify the configuration to “A + D + E + G”. Accordingly, for the first dunning sentence, “reminding” is made by the synthesized voice with the corrected sentence.

  Although the present invention has been described according to the embodiments, the present invention can be variously modified within the scope of the gist thereof. For example, in the case where the notification ID shown in FIG. 2A is “A0300102” and the customer ID is “K120589”, if the attribute “penalty” is “0” or “1”, it is repeated as described above. If the attribute “penalty” is “2” or “3” by correcting the parameter of “1” or correcting the composition of the first dunning sentence to “A + D + E + G”, the call history Even if the record in 222 is “2B”, this may be ignored and the parameter correction as described above may not be performed.

  As described above, according to the present invention, the voice notification device can notify the contents of the dunning or the like by voice having a property corresponding to the age, sex, unmarried / married, etc. of the other party of the call. it can. For example, it is possible to select the voice quality such as voice quality, speed, volume, etc. based on the attributes of the other party of the call and to notify the contents, so that the contents can be provided without causing discomfort to the other party of the call Can be clearly communicated.

  Further, according to the present invention, in the voice notification device, it is possible to make a voice notification based on the contents of the notification according to attributes such as age, sex, unmarried / married, etc. of the other party of the call. For example, it is possible to select the composition of a sentence such as the word order of the sentence, the length of the sentence, the number of repetitions, etc. based on the attribute of the other party of the call and to notify the content, so that the other party of the call is uncomfortable. The contents can be clearly communicated without giving.

It is a voice notification system block diagram. It is audio | voice notification apparatus explanatory drawing. It is audio | voice notification apparatus explanatory drawing. It is audio | voice notification apparatus explanatory drawing. It is a voice notification processing flow. It is a figure explaining the conventional audio | voice notification process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Data management apparatus 2 Voice notification apparatus 3 Telephone 21 Data receiving means for notification 22 Storage means 23 Call content control means 24 Voice synthesis means 25 Voice transmission means 41, 42 Network

Claims (9)

Storage means for storing notification data including a telephone number and customer attributes;
Based on the customer attribute of the notification data, the call content control means for determining the nature of the voice when synthesizing the text that is the call content;
Speech synthesis means for synthesizing the text that is the content of the call into speech of the nature determined by the call content control means;
A voice notification device comprising: voice transmission means for calling the telephone number of the notification data and transmitting the voice synthesized by the voice synthesis means. The voice notification device further includes
Each call has a call history including call state information indicating the end state of the call or the disconnection position of the call,
The voice notification device according to claim 1, wherein the call content control means determines the nature of the voice based on a customer attribute of the notification data and a call state information regarding the telephone number. The voice notification device according to claim 1, wherein the voice property is voice quality, speed, volume, or a combination thereof. Storage means for storing notification data including a telephone number and customer attributes;
Call content control means for determining the composition of the text that is the call content based on the customer attribute of the notification data, and creating the text that is the call content based on this,
Speech synthesis means for synthesizing a sentence that is the content of the call created by the call content control means;
A voice notification device comprising: voice transmission means for calling the telephone number of the notification data and transmitting the voice synthesized by the voice synthesis means. The voice notification device further includes
Each call has a call history including call state information indicating the end state of the call or the disconnection position of the call,
The call content control means determines the structure of the text as the call content based on the customer attribute of the notification data and the call status information for the telephone number, and the text as the call content based on this The voice notification device according to claim 4, wherein: The voice notification device according to claim 4 or 5, wherein the composition of the sentence is the word order of the sentence, the length of the sentence, the number of repetitions, or a combination thereof. The voice notification device according to claim 6, wherein in the composition of the sentence, the composition of the sentence after the second repetition is the longest sentence. The voice notification device according to claim 6, wherein when the other party is an answering machine when connected to the telephone number of the notification data, the number of repetitions in the sentence structure is one. The voice notification device according to claim 1, wherein the customer attribute of the notification data is age, sex, unmarried / married, or a combination thereof.

Images