JP2015049337A - Voice response device, voice response program and voice response method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable suitable response processing according to a customer.SOLUTION: A voice response device comprises a customer identifying unit, an importance determination unit, and a response processing switching unit. The customer identifying unit identifies a customer as an origin. The importance determination unit determines a level of importance of the identified customer. The response processing switching unit switches contents of response processing on the basis of the determined level of importance.

Description

  Embodiments described herein relate generally to a voice response device, a voice response program, and a voice response method.

  In a call center that accepts orders from customers and inquires about products by telephone, voice recognition processing is used to utilize customer voices and reduce the burden on operators. An automatic voice response device of a type equipped with a voice recognition function can recognize a call voice of a customer and can automatically reproduce a response message corresponding to the recognition result.

  By the way, since customers who call the call center make various calls using various communication infrastructures such as mobile phones and IP networks, it is difficult to ensure a certain voice recognition rate. ing.

JP 2001-86243 A Japanese Patent Laid-Open No. 2001-184084 JP 2000-347684 A

  In other words, a customer with a low speech recognition rate of call speech is likely to have a call speech recognition error. When a recognition error actually occurs, for example, until the customer utters a speech and the speech is confirmed as a recognition error. After that, it is switched to, for example, an artificial response process by dialogue with an operator. Therefore, this type of customer results in a decrease in service quality. Here, for important customers with many transactions, we would like to suppress such deterioration in service quality as much as possible.

  Therefore, the problem to be solved by the present invention is to provide a voice response device, a voice response program, and a voice response method capable of realizing an appropriate response process according to a customer.

  The voice response device according to the embodiment includes a customer specifying unit, an importance level determining unit, and a response process switching unit. The customer specifying unit specifies a customer who is a transmission source. The importance level determination unit determines the importance level of the identified customer. The response process switching unit switches the content of the response process based on the determined importance.

The lineblock diagram showing the telephone answering system containing the automatic voice response device concerning a 1st embodiment. The block diagram which shows the structure of the automatic voice response apparatus of FIG. The figure which shows an example of the customer database which the automatic voice response apparatus of FIG. 2 refers. The flowchart which shows the voice response method by the automatic voice response apparatus of FIG. The block diagram which shows the structure of the automatic voice response apparatus which concerns on 2nd Embodiment. The flowchart which shows the voice response method by the automatic voice response apparatus of FIG.

Hereinafter, embodiments will be described with reference to the drawings.
<First Embodiment>
As shown in FIG. 1, a telephone answering system 10 including an automatic voice response device (IVR: Interactive Voice Response) 20 according to the present embodiment includes a public network 11, a private branch exchange 12, an operation terminal 16, a switching hub and an L3 (layer 3) NW (network) switch 15 such as switches, operator telephones 5a, 5b..., Operator PCs 6a, 6b..., CTI (Computer Telephony Integration) device 17, customer DB (customer database) 18, voice recognition dictionary (voice recognition) Database) 19, the above-described automatic voice response device 20 having a voice recognition function, and the automatic voice response device 9 having no voice recognition function.

  The public network 11 is, for example, PSTN (Public Switched Telephone Networks). The private branch exchange 12 is, for example, IP (Internet Protocol) -PBX (Private Branch eXchange). The private branch exchange 12 performs incoming call control and outgoing call control from a customer (user) via the public network 11 in a call center. The operator telephones 5a and 5b, the CTI device 17, the automatic voice response devices 9 and 20, and the private branch exchange 12 are connected to each other via a network 14 such as a LAN and the NW switch 15.

  The operator telephones 5a and 5b are callers for receiving telephone calls from customers (users) and performing telephone response operations (artificial dialogue). The operator PCs 6a and 6b are information terminals for the operator to perform a telephone answering operation while displaying the customer information notified through the NW switch 15 and the information input from the keyboard by the operator on the screen.

  The automatic voice response device 20 and the automatic voice response device 9 reproduce a response message (voice guidance) corresponding to a call received from the private branch exchange 12 side, and respond to a user's request without intervention of an artificial response by an operator or the like. It is a device that responds automatically. Specifically, the automatic voice response apparatus 20 according to the present embodiment having a voice recognition function reproduces a response message corresponding to a voice recognition result of a call voice from a telephone by a customer.

  On the other hand, the automatic voice response device 9 having no voice recognition function reproduces a response message corresponding to a button operation (such as a tone of a selected button number) by a customer from a telephone or the like. Further, in these automatic voice response devices 9 and 20, an operator ID (operator number) which is identification information of an operator who answers the telephone, identification information of the operator telephones 5a and 5b, and the like are set.

  The CTI device 17 receives information (caller number of the customer's telephone, business designation number, etc.) from the private branch exchange 12 side when the operator answers the customer's telephone through the private branch exchange 12, and receives call information (operator ID and business designation number). Etc.) is generated and stored.

  The operation terminal 16 is a terminal operated by a supervisor who is an administrator of the telephone answering system 10. The operation terminal 16 is an apparatus that accesses the private branch exchange 12, the automatic voice response device 20, the automatic voice response device 9, and the like to perform various settings. For example, the operation terminal 16 displays text data obtained by recognizing the voice of a call between the customer and the operator on the monitor, and displays an inappropriate word spoken by the operator during the telephone response process, For example, a keyword search is performed for words necessary for grasping the needs.

  Next, the configuration of the automatic voice response device 20 according to the present embodiment, and the configurations of the voice recognition dictionary 19 and the customer DB 18 will be described in detail with reference to FIGS. As shown in FIG. 2, the automatic voice response device 20 includes a customer identification unit 31, an importance level determination unit 32, a response processing switching unit 33, a threshold setting unit 34, a customer DB storage processing unit 35, a voice recognition unit 37, a recognition rate. A calculation unit 36 is provided.

  Here, the automatic voice response device 20 is equipped with various types of hardware such as a main memory such as a RAM, an auxiliary (secondary) storage device such as an HDD, a CPU, and a ROM. The automatic voice response device 20 loads a voice recognition program stored in advance in an external storage device, a ROM, or the like onto the main memory, so that the customer identification unit 31, the importance level determination unit 32, the response processing switching unit 33, and the like described above are loaded. Each of the above-described constituent elements including is realized by software, for example. Note that these components may be configured by hardware instead of software.

  The voice recognition dictionary 19 is a table that stores voice feature quantities (such as feature vectors) and text data in association with each other. On the other hand, as shown in FIG. 3, the customer DB 18 has identification information such as a customer ID for identifying a customer, a calling number (caller number), name, address, accumulated purchase amount, accumulated number of purchases (or the latest one year). The number of purchases), the voice recognition rate, and the operator ID are stored in association with each other.

  The calling number is a telephone number notified by a calling customer, for example, from a mobile phone or a landline phone used during a call. The cumulative purchase amount is the cumulative amount that the customer has used to date, for example, when purchasing the product sold by the operator of the telephone answering service. The cumulative number of purchases is, for example, the number of purchases made by the customer who has purchased the above-described product or the like (or the number of purchases of the product by the customer in the last one year).

  The voice recognition rate indicates the degree of successful voice recognition for call voices (terms) that the customer has called in the past. Specifically, the voice recognition rate represents the number of times of successful voice recognition of a term, for example, as a percentage with respect to the number of times voice recognition is performed on a term spoken by a customer. For example, a customer who uses a telephone with good call sound quality or a customer who has a good tongue, for example, has a substantially high speech recognition rate. The operator ID is operator identification information. In the customer DB 18, for example, an operator ID of an operator who has received the latest (previous) telephone reception is stored. The customer DB 18 and the voice recognition dictionary 19 may be externally connected to the automatic voice response device 20, or may be mounted on the main body of the automatic voice response device 20.

  As shown in FIG. 2, the voice recognition unit 37 refers to the voice recognition dictionary 19 and converts, for example, a customer's call voice (uttered term) acquired via the NW switch 15 into text data by voice recognition. To do. Here, when the speech recognition unit 37 detects that the probability of the recognition result recognized by speech recognition is equal to or less than the threshold value, the speech recognition unit 37 determines a recognition error and notifies the recognition rate calculation unit 36 of the determination result.

  Further, when the speech recognition unit 37 detects that the probability of the recognition result recognized by speech exceeds the threshold value, the speech recognition unit 37 determines that the recognition is successful, and notifies the recognition rate calculation unit 36 of the determination result. Furthermore, when the speech recognition unit 37 determines that the speech recognition is a recognition error, the speech recognition unit 37 outputs the same response message again, and re-performs speech recognition of the call speech uttered by the customer accordingly. The voice recognition unit 37 repeats such voice recognition re-execution until the number of recognition errors reaches a threshold value or more.

  The recognition rate calculation unit 36 calculates the above-described speech recognition rate based on the determination result indicating the recognition error or the recognition success notified from the speech recognition unit 37. As shown in FIG. 3, the customer DB storage processing unit 35 stores customer identification information and various types of customer information in the customer DB 18 in association with each other. Specifically, the customer DB storage processing unit 35 stores the voice recognition rate and the customer identification information, which are updated each time voice recognition is performed, in the customer DB 18 in association with each other.

  The customer specifying unit 31 specifies a calling customer based on a telephone number notified by a customer, for example, from a mobile phone or a fixed phone. The importance level determination unit 32 determines the importance level of the customer specified by the customer specification unit 31. More specifically, the importance level determination unit 32 refers to the customer DB 18, based on the history of transactions with the customer, that is, the accumulated purchase amount, the accumulated purchase count (or the most recent 1 year purchase count), etc. Determine the importance of the customer.

  The threshold value setting unit 34 is a threshold value that serves as a reference for determining whether or not the voice recognition rate of the call voice (term) that the customer has called in the past is within the allowable range, and the voice recognition unit 37 indicates the voice recognition result. A threshold value used as a criterion for determining the probability, a threshold value used as a criterion for determining the importance of the customer, a threshold value for the number of recognition errors that determines the number of times of speech recognition re-execution, and the like are set.

  Specifically, for example, 90% is set as the threshold value for determining whether or not the speech recognition rate is within the allowable range, and twice or the like is set as the threshold value for the number of recognition errors. Furthermore, the threshold for determining the importance of a customer (whether or not it is a customer with high importance) includes a cumulative purchase amount of 800,000 yen and a cumulative purchase count of 12 (or the number of purchases in the most recent year) As 8 times).

  That is, the customer A illustrated in FIG. 3 has a voice recognition rate that satisfies the threshold. Customer A also satisfies the threshold with respect to the importance of the customer, and is a user with many transactions and high importance. The various threshold values set by the threshold setting unit 34 can be appropriately changed by an input operation from the operation terminal 16 operated by a supervisor or the like.

  As shown in FIG. 2, the response process switching unit 33 switches the content of the response process based on the importance level of the customer determined by the importance level determination unit 32. Here, the response processing switching unit 33 of the present embodiment, based on the determined importance and the voice recognition rate of the call voice that the customer has called in the past (voice recognition rate stored in the customer DB 18), Change the content of response processing. Specifically, the response processing switching unit 33 obtains information obtained from the first response processing for automatically recognizing and automatically responding to the voice of the call made by the customer, and processing different from the voice recognition, and responds by acquiring information from the customer side. And switching to the second response process.

  That is, the first response processing includes a call voice from a telephone by a customer, and a response message (automatic voice guidance) automatically played back on the automatic voice response device 20 side in response to the voice recognition result of the call voice. Response processing to be exchanged. On the other hand, the second response process is a process of responding without using voice recognition. Specifically, as shown in FIG. 1, the second response process includes a process of transferring the customer's call voice (call line) to the operator telephones 5a and 5b (operator call machines). In the transfer process in this case, based on the operator ID stored in the customer DB 18, the customer's call voice may be transferred to the operator's operator telephone that the customer has received the latest (previous) telephone reception.

  Here, the response process switching unit 33 switches to the second response process for a customer whose importance is determined to be greater than or equal to the threshold (customer with many transactions and high importance). In this case, since response processing is performed by dialogue with the operator, for example, detailed requests from the customer side can be dealt with artificially by the operator, and the quality of service can be improved. Further, since the response process does not use voice recognition, waste of extra time due to a voice recognition error or the like can be suppressed.

  Further, the second response process includes, for example, a process of transferring the customer's telephone line to the automatic voice response device 9 having no voice recognition function. The second response process in this case is a response process for exchanging a button operation (such as a tone of the selected button number) by a customer from a telephone or the like and a response message automatically reproduced on the automatic voice response unit 9 side. Become. In this case as well, waste of time due to voice recognition errors and the like can be suppressed, so that the quality of service can be improved for customers with high importance. In addition, the response processing switching unit 33 responds to a customer (a customer with a low voice recognition rate) whose voice recognition rate is determined not to be greater than or equal to the threshold even if the customer is determined to have an importance level lower than the threshold value. Switches to the second response process. Also in this case, it can be avoided that extra time is spent on the response processing.

  Next, a voice recognition method realized by the automatic voice response apparatus 20 configured as described above will be described based on a flowchart shown in FIG. As shown in FIG. 4, when the automatic voice response device 20 detects an incoming call from the private branch exchange 12 (YES in S [Step] 1), the customer specifying unit 31 first determines the customer DB 18 based on the calling number. (S2), the caller customer is specified (S3). At this time, the importance level determination unit 32 determines the importance level of the identified customer. If the determined importance is equal to or greater than the threshold (YES in S4), the response processing switching unit 33 does not respond to the IVR 20 but responds to the customer by the operator 5a. 5b, etc. (S5).

  When the determined importance is less than the threshold (NO in S4), the response process switching unit 33 refers to the customer DB 18 and detects the voice recognition rate of the identified customer's call voice. When the voice recognition rate of the customer's call voice is not greater than or equal to the threshold (NO in S6), the response process switching unit 33 selects the second response process described above for this customer (S5). When the voice recognition rate of the customer's call voice is equal to or greater than the threshold (YES in S6), the response process switching unit 33 switches to the first response process that recognizes the call voice that the customer has called and automatically responds ( The first response process is selected).

  Specifically, when the number of recognition errors in voice recognition is not equal to or greater than the threshold (NO in S7), the response processing switching unit 33 plays a response announcement (automatic voice guidance) (S8), and a customer corresponding to this response announcement. The voice recognition unit 37 executes the voice recognition processing of the call voice (S9), and further reproduces the voice recognition result with automatic voice (S10). Here, when the speech recognition results in a recognition error (NO in S11), the above-described processing from Step 7 (S7) to Step 10 (S10) is executed again.

  On the other hand, if the speech recognition is successful (YES in S11), the recognition rate calculation unit 36 calculates the speech recognition rate by the first response process (S12), and the customer DB storage processing unit 35 The calculation result of the recognition rate is reflected in the customer DB 18 (S13). When the number of recognition errors in voice recognition is equal to or greater than the threshold (YES in S7), the response process switching unit 33 switches the response to the customer to the second response process described above (S5). In this case, in S5, the speech recognition rate is calculated, and the calculation result of the speech recognition rate is reflected in the customer DB 18.

  As described above, in the automatic voice response device 20 of the present embodiment, for a customer who has many transactions and high importance, for example, the response process is switched by interaction with an operator, and so on. It is possible to respond to the above, and the quality of service can be improved. Therefore, according to the automatic voice response device 20, an appropriate response process according to the customer can be realized.

<Second Embodiment>
Next, a second embodiment will be described based on FIGS. In FIG. 5, the same components as those in FIG. 2 described in the first embodiment are given the same reference numerals, and redundant descriptions are omitted. As shown in FIG. 5, the automatic voice response device 50 according to the present embodiment replaces the threshold setting unit 34 and the response processing switching unit 33 included in the automatic voice response device 20 of the first embodiment with a threshold setting. Unit 54 and response processing switching unit 53. The automatic voice response device 50 further includes a recognition rate comparison unit 55, a dictionary update unit 56, and a recognition rate reset unit 57.

  For example, the dictionary updating unit 56 focuses on (specializes) a specific term that is frequently recognized as a speech recognition target, and improves the feature amount (feature vector) of the speech corresponding to the specific term. The contents of the voice recognition dictionary 19 are updated. The recognition rate comparison unit 55 calculates the speech recognition rate calculated by the recognition rate calculation unit 36 before the update of the speech recognition dictionary 19 and the recognition rate calculation unit 36 after the update of the speech recognition dictionary 19 for the identified customer. The obtained speech recognition rates are acquired from the customer DB 18 before and after the update of the speech recognition dictionary 19 and compared with each other.

  When the speech recognition dictionary 19 is updated by the dictionary update unit 56, the recognition rate reset unit 57 stores the speech recognition rate stored in the customer DB 18 before the dictionary update unit 56 is updated in the customer DB storage processing unit 35. To reset. The response process switching unit 53 uses the voice recognition rate before the update of the voice recognition dictionary 19 as a reference, and if the increased rate of the voice recognition rate after the update of the voice recognition dictionary 19 is lower than the threshold, On the other hand, the switching to the second response process that does not use voice recognition is executed permanently.

  In addition to the various threshold values set by the threshold value setting unit 34 of the first embodiment, the threshold value setting unit 54 sets a threshold value corresponding to the increased rate of the voice recognition rate after the update of the voice recognition dictionary 19 described above. Has been. For example, 1% or the like is set as the threshold. That is, the response processing switching unit 53 determines that the improvement effect of the speech recognition rate is hardly obtained when the increase rate of the speech recognition rate after the update of the speech recognition dictionary 19 is, for example, less than 1%. Switching to the second response process is always executed.

  Next, a voice recognition method realized by the automatic voice response apparatus 50 configured as described above will be described based on a flowchart shown in FIG. As shown in FIG. 6, after the customer is specified by the customer specifying unit 31 (S21), the recognition rate comparison unit 55 acquires the speech recognition rate before the update of the speech recognition dictionary 19 for the specified customer. Then, the updated speech recognition rate of the speech recognition dictionary 19 is acquired, and the speech recognition rates are compared with each other (S24).

  If the rate of increase in the speech recognition rate after the update of the speech recognition dictionary 19 is lower than the threshold (YES in S25), the response processing switching unit 53 does not use speech recognition for this identified customer. The switch to the response process 2 is executed permanently. For example, in this case, the customer DB storage processing unit 35 corresponds to the voice recognition rate exemplified in the first embodiment so that the switch to the second response process is always executed for the identified customer. The voice recognition rate (for example, 50%) less than the threshold value (90%) is registered in the customer DB 18 in a non-rewritable state.

  Therefore, according to the automatic voice response apparatus 50 according to the second embodiment, for example, a response process that does not use voice recognition is applied to a customer whose voice quality is difficult to recognize voice or a customer whose tongue is not so good. Therefore, the quality of service for such customers can be improved.

  As mentioned above, although embodiment of this invention was described, this embodiment is shown as an example and is not intending limiting the range of invention. The novel embodiment can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. This embodiment and its modifications are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  For example, as illustrated in FIG. 4 and the like, the first embodiment exemplifies the process of switching the response process based on the importance of the customer and the voice recognition rate of the customer, but the customer's voice recognition rate is considered. Instead, the response process may be switched by a single determination on the importance of the customer. Further, a customer who has reached a specific telephone number (for example, a member-only telephone number) managed by the call center (telephone answering system 10) is determined as a highly important customer (for example, a customer who can receive special treatment), and Thus, the response process switching unit may preferentially switch to the response process corresponding to the operator, for example.

  5a, 5b ... telephone for operator, 6a, 6b ... PC for operator, 9, 20, 50 ... automatic voice response device, 10 ... telephone answering system, 18 ... customer DB, 19 ... voice recognition dictionary, 31 ... customer identification part, 32 ... Importance determination unit, 33, 53 ... Response processing switching unit, 34, 54 ... Threshold setting unit, 35 ... Customer DB storage processing unit, 36 ... Recognition rate calculation unit, 37 ... Voice recognition unit, 55 ... Recognition rate comparison , 56... Dictionary update unit, 57... Recognition rate reset unit.

Claims (9)

A customer identification unit that identifies the customer from whom the call originated;
An importance determination unit for determining the importance of the identified customer;
A response process switching unit that switches the contents of the response process based on the determined importance;
A voice response device comprising: The response process switching unit responds by obtaining information obtained by a process different from the first response process for automatically recognizing and automatically responding to a call voice made by the customer from the customer side and responding. Switching between the second response process and
The voice response device according to claim 1. The second response process includes a process of transferring the call voice of the customer to a telephone that can be handled by an operator.
The voice response device according to claim 1 or 2. The importance determination unit determines the importance of the customer based on a history of transactions with the customer.
The voice response device according to any one of claims 1 to 3. The response process switching unit switches the contents of the response process based on the determined importance and the voice recognition rate of the call voice that the customer has called in the past.
The voice response device according to any one of claims 1 to 4. A voice recognition unit that recognizes a call voice while referring to a voice recognition dictionary that stores voice feature quantities and text data in association with each other;
A recognition rate calculation unit for calculating a speech recognition rate of the call voice by the voice recognition unit;
A dictionary update unit for updating the contents of the speech recognition dictionary;
For the customer, a recognition rate comparison unit that acquires the speech recognition rate calculated before the update of the speech recognition dictionary and the speech recognition rate calculated after the update of the speech recognition dictionary and compares them with each other;
Further comprising
The response processing switching unit, with respect to the speech recognition rate before the update of the speech recognition dictionary, when the rate of increase of the speech recognition rate after the update of the speech recognition dictionary is lower than a threshold, for the customer Permanently switching to the second response process,
The voice response device according to claim 5. A database storage processing unit for storing the speech recognition rate obtained before the update of the speech recognition dictionary and the identification information of the customer in a database in association with each other;
A recognition rate resetting unit for resetting the speech recognition rate stored in the database when the speech recognition dictionary is updated;
The voice response device according to claim 6, further comprising: A customer identification unit that identifies the customer from whom the call originated;
An importance determination unit for determining the importance of the identified customer;
A response processing switching unit that switches the content of the response processing based on the determined importance,
Voice response program that makes the computer function as a computer. Identifying the originating customer,
Determining the importance of the identified customer;
Switching the content of response processing based on the determined importance,
A voice response method.

Images